Guide to Chapter Six of Gareth Evans'
The Varieties of Reference

Rick Grush (Department of Philosophy, UC San Diego)
Copyright ©  Rick Grush 1999, 2000, 2001, 2002
Copyright holder grants permission to distribute, quote, and reproduce for any non-commercial purposes provided authorship and copyright are clearly indicated.

Chapter 6: Demonstrative Identification

Table of Contents:

- 6.1 Demonstrative identification and perception
- 6.2 Information-links are not sufficient
- 6.3 Egocentric spatial thinking: 'here'
- 6.4 Demonstrative identification of material objects
- 6.5 Some Consequences
- 6.6 Immunity to error through misidentification
- Appendix

6.1 Demonstrative identification and perception

This chapter is concerned with demonstratives, and in this section, Evans explores briefly the connection between demonstrative identification and perception. First, he points out a couple of ways in which the notion of perception is vague. Second, he makes some remarks concerning the relation between the theoretical notion of demonstrative identification and the occurrence of demonstratives expressions in natural language. These are prologues to the material in subsequent sections.

The second and third paragraph discuss the relation between 'demonstrative identification' and perception. Demonstrative identification, it will turn out, is a theoretical notion — one of the three kinds of identification of objects (the other two being by description, and recognition-based). A demonstrative identification is a way of knowing what object one is thinking about (and thus suffices, when successful, for the satisfaction of Russell's Principle), a way that makes possible a demonstrative thought. The end of this section will remark a bit on the relation between demonstrative identification (a notion whose application is in understanding certain kinds of thought) and demonstratives (a type of natural language expression), but for now we are sticking to remarks on demonstrative identification of objects and perception of objects.

Footnote 1 raises the objection that it does not seem as though perception is necessary for successfully using demonstratives. The objection (credited to John Wallace) is that one can demonstratively refer to something by pointing with ones eyes closed. Evans' reply, which echoes previous chapters, is that the important notion is one of understanding. Evans has already rightly claimed that one can use a linguistic token meaningfully without oneself understanding that usage (by exploiting a device that has its meaning bestowed socially). But what is important is the ability to understand such a reference. And the claim here is that in order to understand a sentence with a (genuine — see below) demonstrative expression, one must have a certain kind of thought.

Evans remarks that the core of perception is an information link between the subject and the perceived object (the notion of an information link will play a large role in what follows), and then discusses two continua along which the notion of perception varies as one varies the nature of the information link involved. The first has to do with time-lags and what might be called information delays/relays, the second has to do with part-whole relations.

First, the normal sorts of perceptual case in which a subject is perceiving an object in his immediate vicinity via a normal sensory modality shades off as the information channel gets longer, and as there are intermediate media through which the information travels. Evans basically cites examples of cases where it is, and where it is not, felicitous to say 'X is perceiving Y'. These include 'S is perceiving the stars' (long time-lag); 'S sees X on the TV' (circuitous information channel); we have 'S sees X in the mirror' (where X is casting a reflection) but not 'S sees X on the ground' (where X is casting a shadow); we can say 'S hears X on the radio' but not 'S hears X on the phonograph'.

The second continuum concerns part-whole relations, and specifically, the ability or inability to demonstratively identify something when one is completely 'aware of' only a part (or perhaps: when one can perceive only a part) of it. Examples include the ability to demonstratively identify and island when perceiving a part of the beach; to identify a chair or other physical object when one is perceiving (or better: sensing) only a surface; to identify a city when one is in a room in the city; etc. These are simply brought up as examples of how the notions of perception and demonstrative identification are unclear and stand in need of regimentation.

The final paragraph broaches an interesting topic: the relation between demonstrative identification as a theoretical concept in the semantics of thought, so to speak, and the occurrence of demonstrative expressions such as 'this' and 'that' in natural language. The question is whether one should take for granted at the outset that whenever a demonstrative expression is being used, that the same semantic natural kind, so to speak, relying on thoughts involving demonstrative identification is being invoked; or if it rather might be the case that there is a semantic natural kind — demonstrative identification — that is often but not always mirrored grammatically by a 'demonstrative' expression. Evans opts for the second. The example given is one of so-called 'deferred ostention', in which, e.g., one points to a car burdened with parking tickets, and says (using a demonstrative expression) 'That man is going to be sorry.' Evans claims that though this sentence has the grammatical form of a demonstrative (a demonstrative expression is in the subject position), it is semantically a description — something like 'The man who owns this car'.

So since surface features of grammar cannot be our guide, we must look rather to the relation between perception and thought. I take it that the idea here is that if we understand how perception of an object can make a certain kind of thought possible, we will have the tools needed to understand the theoretical notion of demonstrative identification. A (true) linguistic demonstrative will be an expression-type such that in order to understand the expression, one must have the sort of thought that is made possible through a demonstrative identification of the object in thought.

6.2 Information-links are not sufficient

There are four strands to this section. First, we have a characterization of information-links (ILs). Second, some considerations aimed at showing that ILs are not by themselves sufficient to sustain demonstrative thoughts (though they are necessary). Third, thoughts about what else, besides an IL, is required to sustain such thought. Finally (the last paragraph), Evans claims that cases of identification involving long time-lags and such appear to be sui-generis, and neither genuinely demonstrative nor descriptive.

First we get a characterization of ILs and their role in thought. The subject S has a controlling conception of the object that is being fed information about the object by an IL. This information is mainlined, so to speak, directly into this dossier (information that makes it into the dossier by means other than mainlining will be discussed in 6.6). The subject then treats this information as germane to the assessment of thoughts about the object. For example, information coming in through the IL to the effect that redness is instantiated is treated as immediately germane to the thought that the object is red. Having such a dossier fed by an active information-link is a necessary condition for a demonstrative thought (though as we shall see there need not be any actual information coming in through the link, as in the cases of 'here' and 'I'). Having the capacity for such an active information-based thought relies on an ability to stay selectively sensitive to information coming from a single object, as when one pays attention to a single violin in a quintet. Without such an ability, there would be no possibility of our maintaining distinct information links.

Second, Evans provides an argument to the effect that an information link is not sufficient, by itself, to sustain demonstrative thought. The brief version of the argument is that a constraint on thought is the Generality Constraint, and that merely having an active information link with an object does not guarantee satisfaction of the GC. Less briefly: In order to have a thought about an object, one must have an adequate Idea of the object — an adequate Idea being what allows the subject's thought to satisfy the GC, an Idea combinable with all allowable predicates. Evans has argued that there are two ways for a subject to have an adequate Idea of an object. Either one's Idea can involve a fundamental identification, or one would have to know what would make an identity proposition between one's idea and a fundamental identification true. It will be easier to see why Evans thinks that a mere information link is not sufficient for this by looking first at what Evans thinks does allow one to have an adequate idea in the case of demonstrative identification.

This brings us to the third topic. In the case of demonstrative identification that is working, not only does on have an information link, but this link allows one to locate the object is space. Recall that for material objects, location in space at a time is the fundamental ground of difference. Having an Idea of an object as the occupier of a spatial location at a given time is a fundamental identification of such an object. With demonstrative identification that works, there are two possibilities. First, one's information link with the object is such that one just knows where the object is in egocentric space — one can see it directly in front of one, for example. This would be a fundamental identification of the object. Or, the information link, while not immediately allowing one to place the object in egocentric space, is such that the subject could use the information link to locate the object. If the information link has this character, then the subject will know what it is for his Idea of the object to be identical to a fundamental identification of a material object, it is his ability, or practical capacity, to locate the object in egocentric space, even if he does not at the time know exactly where it is. (This paragraph has been a quick gloss, that leaves out a number of central themes, on an account that will be spelled out more fully in 6.3.)

The mere existence of an information link, however, need not allow the subject to locate the object in space (or better: need not be such as to allow the subject to engage any practical capacities to locate the object in space). In such a case, the subject will not have an adequate Idea of the object, and hence will be unable to think thoughts about it (at least unable to think demonstrative thoughts — the subject of course could still think about the object by description, as the object that is the source of a given body of information). I will make a few more remarks about this argument in just a moment.

Fourth and finally, Evans comments that certain sorts of case, those involving large time-lags, seem to be neither demonstrative, nor descriptive, but sui generis. There is little argument or discussion here, and nothing hangs on this point for later purposes, so I will drop it.

Evans' argument here will probably strike many as unconvincing. I will try to make it as convincing as possible: this will involve a few steps. First, one can plausibly take it that, for example, one's grasp of certain concepts is manifested in an ability to decide their applicability. For example, suppose that while in a paint store one day, I overhear someone saying they want teal paint. I infer correctly that teal is a shade of color, but do I have the concept of teal? It seems plausible to say that I simply don't. If given a bunch of color chips, in conditions as favorable as one likes, I couldn't tell you which were teal and which not, nor even come close, then I simply don't know what teal means — I don't have the concept teal. I might repeat the words 'The house is teal, the cat is teal, the box is teal' to myself, but I would not be able to think the thought 'the house is teal'. (Of course, I can think the thought the house is some color or other, and if I know that I don't know what 'teal' means but know that it is a shade of color, and take 'teal' as a place-holder for 'some color or other', then I can think this other thought. But thinking that the house is some color or other is not the same as thinking that the house is teal. I think many who find this argument unconvincing are allowing themselves, amazingly, to slide from some color or other to teal as though this were not an unacceptable thing to do.) The next step in the reconstructed argument is to take an example similar to Evans' example of the soccer player on the television, but which is more convincing (this example is due to Adrian Cussins). Imagine a television with an image of an assembly line along which hundreds of Coke cans are moving along. The subject points to one (or perhaps merely fixes his attention on the image of one) and wonders whether than can will be purchased by a movie star. Part of this wondering is the entertaining of the thought 'that can will be purchased by a movie star'. In such a case, the subject is in a position with respect to that coke can that the subject in the teal example was with respect to teal. The subject simply has no idea what it would mean for that can, rather than any other can of coke, to be purchased by a movie star, he has no idea of that can that distinguishes it from any of hundreds or thousands of other coke cans. Now again, there are thoughts similar to the attempted demonstrative thought that the subject can very well entertain (and it is this fact that, I think, makes people more willing than they should be to credit the subject with an ability to entertain a demonstrative thought about that coke can). For example, the subject can think that some coke can or other will be purchased by a movie star (an existential coke can thought), and the subject can think the descriptive thought that the coke can that is producing this image will be purchased by a movie star. But though there are relations between the existential thought and the descriptive thought, they do not amount, individually or jointly, the purported demonstrative thought.

[Note: Footnote 10 says that 'we have made a slight alteration to our conception of information-based thoughts', and goes on to say that while in Chapter 5 a thought was information-based just in case its controlling conception contained information from the object, it now appears that we are taking an information-based thought to be one that is relying on a certain information link. This might be misleading. Demonstrative thoughts will be information-based and also will rely on a specific information link. But Evans will still want to use the expression 'information-based thought' in the original sense later on.]

6.3 Egocentric spatial thinking: 'here'

Section 6.1 posed the question How does perception make a thought possible? We will now get an answer to this. Perception will make a thought possible by allowing the subject to place the perceived object at a location in space at a time. In so doing, perception provides a fundamental identification of the object, and this makes thought about it possible. This section spells out how this happens — how perception allows us to locate something in space. This section is one of the longest in the book and could easily have been broken into two or more. Six topics can be discerned. First, a gloss on what is needed for a fundamental identification of a material object — the ability to make such an identification being crucial for demonstrative identification. This turns out to be objective space. Second, there is a long discussion of egocentric spatial representation. The upshot is that an experiential episode carries egocentric spatial content if that experience is hooked up in the right way with the complex web of sensation-action connections that a creature has. Thus, in the normal case, perception of an object makes its location in egocentric space manifest. Third, there is a discussion of our ability to bring egocentric space into coordination with objective space, thus allowing us to locate perceived objects in objective space and hence to have a fundamental identification of them. Fourth, there is a discussion of 'here' aimed at showing that even 'here' thoughts require more than an information link. Fifth, a brief comparison between the demonstrative identification of places and objects. And sixth, an argument that demonstrative thoughts about places are Russellian. In order to try to keep things manageable and clear, I will number these six topics in this section as 6.3.1 - 6.3.6, and treat them more or less separately.

6.3.1. Introduction (top of p. 151 to bottom third of p. 152, and a recap from middle of p. 168 to bottom of p. 168)

The fundamental ground of difference of physical objects is their spatial location at a time, so in order to have any adequate Idea of an object for use in demonstrative identification, we must either be able to locate the object (or location, if we are attempting a thought about a place), or know what would make such an identification of its location with a location in objective space true. Either way, what is crucial is the ability to represent objective spatial locations. We represent (objective) space by means of a cognitive map: a representation of the spatial relations between the objects in the world that we frequent as well as the large scale landmarks that establish their objective location. So, in order to represent something as being at a location in space at a time, we must represent it in our cognitive map of the single spatio-temporal world, and this involves representing it as being spatially related to other objects at places that constitute the framework for identifying such places. But the sort of spatial representation involved must be objective — it cannot be from a point of view, for locations-from-a-point-of-view, or egocentric spatial locations, do not provide fundamental identifications of objects. This is because knowing where something is in relation to me does not necessarily tell me where the object is located in objective space, and thus does not serve to distinguish it from all other things of its type. Though as we shall see, knowing the location of something in egocentric space is the first step towards knowing where it is in objective space. If something is located in egocentric space, and if in addition the subject has the capacity to align her egocentric space with an objective cognitive map, then the subject will ipso facto be able to locate the perceived object in objective space, and will have an adequate Idea of it.  It is to the notion of egocentric space that we turn first.

6.3.2. Egocentric space (bottom third of p. 152 to the top of p. 162)

The basic idea of this subsection is that an ability to represent egocentric space is made possible by the complicated web of behavioral dispositions manifested in sensorimotor skills. A better way of putting it is as follows: an experience E has egocentric spatial import for a creature if that experience automatically (noninferentially) keys spatial-behavioral dispositions with respect to it. Now to some more detail.

Evans first turns to a proposal that 'here' is thought of by description as 'the place I occupy'. Evans rejects this, remarking that i) part of what makes a place 'here' is that it is the area within which perception and action are coordinated — the space of behavior, so to speak — and this might come apart from 'the place I occupy' (this is, more or less, the import of the brain in a vat example on page 153); and ii) that 'I' does not have priority over 'here' — Evans' point here seems to be that indexicals such as 'I', 'here' and 'now' are inter-defined: they form a local holism. One does not define some of them on the basis of the others, but rather the three are given together and inter-define each other.

[Note: Evans' remarks on the brain in a vat example could be made more strongly. Imagine someone in the situation described in The Matrix, who, while still in the virtual reality program, is appraised of her situation. She might come to believe (truly) that this thing between what she takes to be her ears — what she would have identified as her brain before being appraised of her situation — is not the thing sustaining her thoughts. She might be able to think that there is such a thing, somewhere (though it is not anywhere in the universe as she knows it) there is something that sustains her thoughts. But it is pretty clear that thinking of such a place as 'here', in the same way that I might think of my bedroom as 'here' when I wake up in the darkness, is out of the question. The relation of this line of thought to the claim that here-thoughts are not descriptions of the form'where I am' is not entirely direct, since 'where I am' and 'where the organ sustaining my thoughts is' are not the same. But the point that  here-thoughts are constitutively linked to possibilities of sensation and action should be clear enough.]

Evans then claims that 'here' is part of a cluster of related spatial notions, including 'there', 'to the left', etc. While it is the case that 'here' thoughts get a grip via connections to perception and action, the details of the sorts of perception and action will mark the difference between locations in egocentric space. The position is summed up nicely in the first full paragraph on page 154:

[Note: I should like to emphasize that Evans here is not endorsing the 'motor theory' of spatial perception, whose proponents include Berkeley, Mach and Poincare. Evans explicitly argues against the motor theory. Spatial content is not stated in muscular terms, or in terms of exertion of proprioception. Rather, it is its own sui generis representational scheme constructed from coordinated sensory or motor interactions and possibilities. See my 'Self, world and space' for more detail on this proposal. Evans repudiates the motor theory at the bottom of page 156 and the top of p. 157.]

The last sentence of this quoted passage is important. Evans is not denying that a subject might be able to respond differentially to stimuli that are in fact correlated with the spatial location of those stimuli. The point is that such perceptions will be taken by the subject to have spatial import only if they are woven correctly into the subject's sensorimotor skills. This idea is spelled out over the next few pages of the text.

At the bottom of page 157, and continuing to the middle of p. 159, Evans turns to arguments to the effect that having an egocentric space, and having the ability to perceive things in egocentric space, does not yet provide for a notion of experience (and I think that Evans means to be using 'experience' in the thick, Kantian sense of 'experience of an objective world'). Merely having the machinery required for an egocentric space does not give one experience of an objective world, nor is it experience for a subject.

To make that next step, something else is required. Evans considers and rejects a proposal that what is required are metathoughts, thoughts that self-ascribe such egocentric 'experiences'. Rather, Evans argues that it is not that the candidate subject have metathoughts, but rather that he have some thoughts, thoughts about the objective world. This will happen for a being that has not only the sort of egocentric representational machinery described, but also has a concept-applying, reasoning system, into which the first sort of mechanisms feed. This reasoning superstructure would allow one to make judgments based upon the deliverances of the infrastructure. (This discussion is very important for the notion of nonconceptual content: a topic that has gained much momentum and sparked much debate since Evans' death. Since this topic is not central to the ideas of this book, I won't elaborate on them.)

Evans returns to the main line of argument — the egocentric spatial content of perception — at the middle of page 159. He there remarks that the sort of content in question concerns a number of simultaneous locations or objects; it provides one with a single, unified representation of a number of simultaneously existing locations. (This fact will be of importance later, when the topic of aligning one's egocentric space with a cognitive map of objective space comes up. It is also an issue that is closely connected to his papers 'Things without the mind', and 'Molyneux's question'.)

Finally, Evans turns to the following important and interesting point, which is that it is possible (though more difficult) to maintain the sort of behavioral (sensory and action-oriented) dispositions with respect to a place or object even in absence of input from it. It is not that the possibilities of action and perception are completely missing. They are there, but they are merely potential. But so long as these potential patterns of dispositions are maintained, they can provide for the sort of spatial content in question. For example, I can have a thought about a particular place that I am not currently receiving any information from, such as a spot just behind my head, provided I am attentionally tuned to any information that might be forthcoming from that location — part of maintaining a complex dispositional connection to the place. In such a case, one might say that I am maintaining an information link with that location even though no information is being passed through that link. (This is the import of footnote 20.)

6.3.3. Objective space (top of p. 162 to middle of p. 164)

The discussion of egocentric space is only a preliminary, for being able to locate some place or object in egocentric space does not give one a fundamental identification of it. A fundamental identification of a material body (or a place) places it at a position at a time, and the positions and times must be objective positions and times in the unified world order. Representing something as being in front of me now does not by itself constitute locating it in this stronger objective sense. So how do we do it? Evans takes it for granted that we have the ability to form a cognitive map of the world — we do not represent the entire universe of course, but anything in the universe is representable on it. What exactly the cognitive map is is unclear, but it is a representation of space; it is the representation that we are unable to specifically represent ourselves in when we are lost. That there is such a thing that is separate from our representation of egocentric space is clear enough. It is why 'I am here' does not count as an answer to the question "Where am I?" when I am lost. Knowing where something is in egocentric space — including knowing that I am at the center of my egocentric space — does not guarantee knowing where it is.

But we have the ability to coordinate (Evans' expression is 'effect a coincidence between') our egocentric space and our cognitive map, and when we do, we thereby locate the things in our egocentric space in objective space. When such a coordination is made, we thereby have made a number of identity propositions of the form [pi = p], where p is some place in egocentric space, and pi is a place in objective space — a fundamental identification of a place.

Now on Evans' account, one does not in fact have to have one's egocentric space coordinated with a cognitive map in order for one to think objectively about locations in egocentric space. What one needs is the ability to so coordinate. If the spaces are coordinated, then one is in fact identifying a place via a fundamental identification. If one is not at the time coordinating them, then provided that one has the ability to do so, one at least knows what it would mean for such an identity proposition to be true. Either way, one has an adequate Idea p of the place. (Recall that an adequate Idea just is, in Evans' terminology, an Idea that is either fundamental, or is such that one knows what an identity proposition between it and a fundamental Idea would consist in.)

6.3.4. 'Here' (middle of p. 164 to the middle of p. 168)

This subsection is aimed at showing that information-links with a place are not sufficient for thoughts about that place, because they may not allow the subject to locate the location in objective space. The example constructed is that of a person who is remotely operating a submarine, and whose sensory input and motor output are as closely linked to the sensors (video cameras, etc.), and effectors (propulsion, excavators, etc.) as one likes. As Evans points out, it might be possible for the subject to throw oneself into the situation, so to speak — paying less and less attention to what is happening in the vicinity of his biological body and being sucked into the situation of the submarine. Let us suppose that the subject sees a rock and wonders 'where is this rock?' There are three possibilities to consider.

First, the subject may have no idea where the sub is (and no way of determining its location), and hence no idea where, in relation to his biological body, the 'seen' rock is. The rock could be 'identified' in a sort of mock egocentric way, but this would not answer the question where the rock is. Alternately, one might issue thoughts about it like 'it is the rock that is in front of the submarine that I am informationally linked to' and thus identify it by description; this would not be a demonstrative identification. But the normal means of identifying something perceptually and knowing where it is in space are not in play.

Second, one may know where the submarine is, and think of it as, say, about a mile out that way and straight down. One can then more or less locate the rock in one's egocentric space, and hence in objective space. But this is because one has located the submarine in egocentric space. One has a fundamental idea of the rock, because one can locate it in one's real egocentric space. The information link by itself is not doing the work.

The third possibility is that one has a cognitive map that one coordinates with the 'egocentric' space of the submarine. This is the sort of case Evans discusses on the central paragraph of page 167. If one does this, then one can know where the rock is, but note that more is going on here than just an information link. It is an information link together with the subject's cognitive map and self-locating skills that are making the fundamental identification of the rock possible.

The point of this section is that merely having an information link does not guarantee that one will be able to, even hypothetically, locate the object or place that is the source of the information in objective space, and hence does not guarantee an ability to have an adequate Idea of it, and hence does not guarantee the ability to entertain thoughts about it.

6.3.5. Demonstrative identification of places vs. objects (top of p. 169 to bottom third of p. 169)

This subsection is brief. Evans is merely comparing demonstrative thoughts of places with those of things. The commonalities are that each requires at least that one maintain a stable disposition to receive information about (and maintain some set of dispositions toward) the demonstratively identified place or thing. Often these dispositions are actualized and one is in fact getting information from or behaving towards the place or thing, but maintaining the dispositions even if their execution is latent is what is key.

The difference is that it is typically much easier to maintain such stable dispositions with respect to places — at least large places, and especially 'here'. In order to maintain them with respect to objects, we typically need to stay in informational contact with the object (as it moves, or as we move) in order to maintain the right set of dispositions. (But Evans will remark later that when it comes to finer distinctions in egocentric space, maintaining stable dispositions with respect to them is very difficult, and is usually pulled off by maintaining a disposition with some object at the place or in some stable spatial relationship to the place.)

6.3.6. Demonstrative thoughts about locations in egocentric space are Russellian (bottom third of p. 169 to end of section)

Evans here argues that the sort of thoughts discussed in this section — thoughts about egocentric locations, especially 'here' —are Russellian. The first step in the argument is to point out that they are not descriptive (Evans here is at least in part, no doubt, in agreement with Perry's points in 'The Problem of the Essential Indexical').

The argument for the Russellian status is in the final paragraph of the section. Here is what he says, in full:

I have to admit that I find some difficulty trying to excogitate the argument out of this passage. But here is one try. A lot hinges on what a thought is. We can follow Frege, as surely Evans does, in saying that a thought is that for which the question of truth arises, and the content of the thought is the condition under which it is true —its truth condition. Now let us flesh out Evans' example: a blindfolded subject is on a very quiet hospital gurney, and is moving down a hallway without realizing it: he takes himself to be stationary. Let us also imagine that the hallway is full of people who are mostly quiet, but by coincidence as the gurney approaches people they begin chatting, and as it passes and moves from them, they stop talking. To the subject on the gurney it sounds as though a continual conversational din is surrounding him, when in fact a wave of chatter happens to be pacing the gurney in an otherwise quiet hallway. The subject, believing himself to be immobile, says 'It is quite noisy here.'

We can assume that in fact there is not any one location toward which the subject has maintained a stable disposition. Now there are two possibilities. First, we can formulate a condition that would provide for a genuine content and truth condition, such as 'There was noise in the vicinity of the gurney', which is a clear truth condition, and would even be true in the circumstances envisioned. If this were the thought essayed by the subject, there would be a thought with a definite content, and no problem. But by hypothesis, the subject is not thinking any description, so such a description as this is not capturing the content of the attempted here-thought.

On the other hand, if there were a place with which the subject maintains stable informational and behavioral dispositions, we could also provide truth conditions: the thought would be true if that place was noisy. But in the given case this does not work. There is no actual place to provide the content of the thought, and there is no description that reflects the content of the attempted thought. So we are left with the conclusion that the thought attempt, because of the sort of attempt it is (non-descriptive) is such that, in the absence of an object (or stable location), there is no specifiable truth condition that reflects the content of the attempted thought, and hence there is no content, and hence no thought. Of course, there are a number of very similar thoughts that the subject might have thought, and even thought truly, such as the descriptive one. The fact that it is so easy to slide back and forth between these related but distinct thoughts perhaps helps to explain why it seems difficult for many to find Evans' argument here convincing.

6.4 Demonstrative identification of material objects

After nicely summarizing the view on demonstrative identification, Evans explores two main threads in this section. The first is a discussion of the relation between identification of places and identification of things, in particular he argues that our idea of a demonstratively identified object is not a description based on a prior identification of a place, as 'the object at p' — our ability to represent locations is not more basic than our ability to represent objects. Second, he discusses the essential temporal extension of demonstrative thoughts, their essential exploitation of abilities to keep track of objects over time. Between these two discussions, Evans provides an argument to the effect that demonstrative thoughts about material objects are Russellian (this argument is parallel to that at the end of 6.3). A few brief programmatic discussions round out the section.

Evans begins by summarizing the theory of demonstrative identification of material objects:

In other words, in the normal case, one's information link with the object allows one to locate it in egocentric space (at p, where p is an Idea of a location in egocentric space). Given that one has an ability to effect a coincidence between one's egocentric space and objective space, one will ipso facto be able to identify every position in egocentric space p with a location in objective space pi, and will therefore be able to place the object at an objective location — one will have made a fundamental identification of the object as the occupier of a location in space at a time. That is, the information link allows one to locate the object of the demonstrative thought, and thus provides for an adequate Idea of the object. In addition to the fact that the information link provides for an adequate Idea of the object, it also supplies the actual information that informs the subject's controlling conception of the object.

One issue worthy of further thought is this: the process by which the information link provides for an adequate Idea of the object has two stages. The first is the stage at which the object is located in the egocentric space. The second is the process by which the egocentric space is brought into coordination with an objective cognitive map. Either or both of these stages might be potential rather than actual, and yet still allow for an adequate Idea. As to the first stage, Evans allows not only cases where the information link immediately locates the object in egocentric space to underwrite adequate Ideas, but also cases in which the information link puts one in a position to locate the object, even if one does not immediately do so. His examples are the sound of the beetle, or the fish tugging on the line. Similarly the second stage can be satisfied by the correct abilities even if their adequate execution is temporarily blocked. So when one is lost, and temporarily unable to effect a coincidence between one's egocentric space and an objective cognitive map, so long as one has the ability to do so, Evans says that one still has an adequate Idea, since one knows what would make the identity between egocentric and objective locations true.

The bulk of page 171 covers points made elsewhere — that information links are not sufficient, and that the notion of understanding an utterance is fundamental.

From the bottom of page 171 to the top of page 173 Evans discusses the relation between a genuine demonstrative identification of an object and a descriptive identification of an object as 'the object at p', which rests on a prior identification of a place. Evans' first argument against this proposal is that, unlike his own proposal, it does not seem able to account for cases such as the beetle, whose location we are not immediately in a position to know. Second, we seem able to demonstratively identify objects when, because the object is moving so quickly, we are unable at any moment to assign it a definite location (imagine a fly quickly buzzing about one's head, before one can latch on to a determinate location, the fly has moved on). His final argument is that we seem to be able to make more demonstrative object discriminations than we are able to make reliable spatial discriminations. A given pill-sized region of a billiard table is not the sort of thing that we can reliably re-identify or even maintain a firm hold on, when paying attention, over a span of time; whereas an actual pill at that location, even if surrounded by many similarly shaped and colored pills, is easy to maintain attention on. Indeed, this sort of case seems to suggests that in some cases at least, it is the location of places that is dependent on a prior identification of objects.

The bottom of page 173 to the top of page 174 is an argument that demonstrative thoughts about material bodies are Russellian. This argument parallels the one at the end of 6.3 (which concerned 'here' thoughts) in structure. The argument structure depends on the premise that demonstrative ID of material bodies is not by description, and two different suggestions to the effect that such identification is by description are discussed and rejected. The first (discussed from the bottom of page 171 and continuing to the top of page 173) is the suggestion that a demonstrative identification of an object is a description of the form 'the object at p'. I have already discussed this case. The second (discussed in footnote 44) is the suggestion that it is a description of the form 'the object I now have my gaze fixed on'. The argument against this proposal is densely packed into the footnote. The structure of the argument is this: Evans claims that it is that fact that my gaze is fixed on O that determines it as the object of my thought, not the 'idea' of my gaze being fixed on O. By this he simply means that no description of the form 'The one I have my gaze fixed on' is in play. The revolves around the claim that the two proposals (Evans' and the descriptive proposal) give different say different things concerning the question 'Is this the one I have my gazed fixed upon?'. On the descriptivist account, this question not only can be meaningfully raised, but it also cannot be answered; for if the mere fact that I have my gazed fixed on O does not entail that I am thinking of O, then it is possible that I have my gaze fixed on objects I am not thinking of. On the other hand, on Evans' proposal, the question 'Is this the one I have my gaze fixed on?' cannot be meaningfully raised. The reason why it cannot be meaningfully raised will be explored more in section 6.6. But for now what is important is the following,. The descriptivist has a sort of explanation of this fact, for on the descriptivist account, the question might be thought to reduce to 'Is the object I have my gaze fixed on the object I have my gaze fixed on?'. But this won't work here, because on the current side of the dilemma we are assuming that the mere fact of the gaze is sufficient to establish that O is the object of thought, thus the descriptivist formulation of the question above is not necessary.

From the top of page 174 to the top of page 176, Evans discusses the temporally extended character of demonstrative identification. This is material that is also treated in the first section of the Appendix to this chapter, Section 6.A1. He first points out a third contribution made by the information link, in addition to the two mentioned above (I am listing these in an order different from that in which Evans lists them at the top of p. 174), to the effect that i) it allows the subject to locate the object in egocentric space, and thus is an enabling condition for having an adequate Idea of the object; and ii) it is the source of the information that goes into the controlling conception. The third contribution mentioned is that the information link provides for a temporally continuous stream of information from the object.

The point here seems to be that a demonstrative Idea appears to be one that has a certain kind of temporal thickness. The argument here is that the thought 'This will be F (in a fraction of a second)', for example 'This pencil will drop (in a fraction of a second)',thought when one sees a pencil rolling toward the edge of a table, is a thought that is confirmed when, in a fraction of a second, the pencil drops (or disconfirmed if it does not). The relationship between this later information and the current thought cannot be analyzed in terms of an identification of the current demonstratively identified object with a future demonstratively identified object, for in fact no such articulation appears to be possible. The same demonstrative Idea spans the temporal gap, so to speak. Given this, demonstrative Ideas must inherently be able to exploit continuing information links of the sort capable of delivering a temporally extended stream of information. Crucial aspects of this argument will be explained in more detail later in the chapter.

The final paragraph of this section is not directly tied to any crucial inferential moves in the book, but expresses an interesting kind of conviction that Evans' takes over from his teacher Peter Strawson. Strawson, in Individuals and The Bounds of Sense, articulated a sort of analysis of the conceptual scheme of our ordinary thought, and this scheme had, according to Strawson, a number of components such as the subject having a bodily presence in the world among other objects, of tracing out a continuous path within them, and of being able to holistically re-identify objects and places. If Evans had had more time, it is likely that he would have treated these topics in more detail. 

6.5 Some consequences

This section touches on three consequences of the view as presented so far. The first concerns the limits of demonstrative identification. The second has to do with the role that sortals play in demonstrative identification. The third has to do with kinds of error that are compatible with successful demonstrative identification.

The first consequence discussed by Evans is concerns what the theory as presented so far says about the boundaries of demonstrative identification, questions that were raised in 6.1. The first kind of case mentioned concerns part-whole relations. Evans points out that because his account relies on the notion of an information link, rather than the more philosophically unclear notion of perception (perception being one kind of information link), cases that puzzled Moore become less puzzling. I often have an information link to an entire object even in cases where in some overly strict sense I might be said to be perceiving only part of it. Examples of the second sort of case involve attempted demonstrative identifications of a city or house when sitting in a closed-up room. Evans claims that since one is not in informational contact with a city even when one is within the city if one is sitting in a closed-up room, one cannot strictly demonstratively identify it as 'this city', attempts to think of the city would have to be descriptive, like 'the city in which I am now situated'. One might wonder what the relationship is between 'this city' and 'here', since Evans allows that one can have 'here' thoughts even in absence of any actual information. In that case, Evans claims that being disposed so as to act on any information should any arise is sufficient for having a 'here' thought. It is not clear what the difference amounts to. It is not clear why the lack of information impugns 'this city' thoughts but not 'here' thoughts. It is possible that the next two topics — sortals and kinds of allowable error — were initially raised here as Evans was writing because he saw them as offering an answer to this question, even though no trace of this account is present in the published version. I will return to a few speculations as to what this might have been in my discussion of  part 2 of the Appendix, which also discusses the role of sortals.

The second issue has to do with sortals, and raises a third respect in which an Idea may be adequate though not fundamental. The first two, which I mentioned in the discussion of 6.4, are that one's information link may not immediately locate the object in perception (the sound of the beetle), but the Idea is adequate so long as one has the capacity to locate it in egocentric space on the basis of he information link; secondly, one may be lost, and hence not able to place the object located in egocentric space in objective space, but the Idea is adequate so long as one has the capacity to figure out where one is, and effect a coincidence between egocentric and objective space. This third way is as follows: in Section 4.4 where Evans introduced and explained fundamental Ideas, it was claimed that fundamental ideas of physical objects involve location in space at a time and a sortal. We now are told that the Idea can be adequate (though not fundamental) so long as there is a way of finding out what kind of thing one is identifying demonstratively.

The third issue concerns another kind of potential error. When one thinks 'this G is F' one can be mistaken not only about the fact that it is a G (one may have the wrong sortal, as discussed above), one might also be mistaken about F. The most obvious application of this is in Donnellan type cases, where one sees a woman in the corner drinking water and says 'the man in the corner is drinking a martini'. Evans here allows that the demonstrative identification is adequate, even though one is mistaken both about the sortal and the attributed property.

6.6 Immunity to error through misidentification

This section brings up an issue that is a result of the kind of analysis Evans has provided of 'this'-Ideas and 'here'-Ideas: the fact that they give rise to judgments that have the peculiar property that they are identification-independent or immune to error through misidentification. Reasons for the inclusion of this discussion here would seem to be a) the phenomenon is a recognized and important one, and it is not insignificant that Evans' analysis of 'this'- and 'here'-Ideas predicts and accounts for the phenomenon. Second, the phenomenon will be playing a significant role in the next chapter, and setting it up here correctly will aid that discussion.

From the bottom of p. 179 to the middle of p. 181 Evans has an initial discussion of the difference between two kinds of 'knowledge'. Though this discussion is often put in term of 'knowledge', the cases discussed need not involve veridicality, and hence 'judgments' or even 'thoughts' might be better. I will use the expression 'thoughts' in my discussion. The point will be to make a distinction between two kinds of thoughts, those that are identification-dependent and those that are identification-free. Starting with identification dependent thoughts: a thought that 'a is F' is identification-dependent if it can be resolved into two thoughts of the form 'b is F' and 'a = b'. So for example, when I see someone I take to be an acquaintance named 'Bob' and think 'Bob is sunburnt', my thought can be seen to result from two distinct thoughts: 'this person is sunburnt'; and 'this person is Bob'. Such a thought might be false not because I have mistaken the blush for a sunburn, but because the identity does not hold: the person I am looking at is not in fact Bob.

By way of contrast, the Ideas involved in 'here'-thoughts and 'this'-thoughts are Ideas that are constitutively tied to certain ways of gaining information — that is, certain information links. A 'here'-Idea is one that a subject has in virtue of mainlining information gained about his immediate vicinity to the controlling conception of his 'here'-thoughts; and a 'this'-Idea is one such that the information link with the object demonstratively identified is mainlined directly into the controlling conception of that Idea. One way to put this is that when either of these Ideas is involved, it is tied to one or more information links in such a way that information to the effect that a predicate is instantiated is mainlined directly into the controlling conception of that Idea, with the result that the subject to have information to the effect that the object in question has that property. Thus, such thoughts do not involve an identification component, and hence are identification-free. My thought that 'here it is dark' cannot be resolved into two thoughts of the form 'it is dark at location p' and 'p is here'. Rather, when the information links that are constitutive of my 'here'-Idea, most relevantly information I gather through vision, register a paucity of illumination — that is, when they carry information to the effect that the predicate '... is dark' is instantiated — this information is mainlined directly into my 'here'-Idea in such a way that I am disposed to judge that 'here is dark'. By way of contrast, if I am watching a video monitor that is displaying a feed from a video camera that I know to be trained on location q, and the information on the monitor makes it manifest that it is dark at location q, then, supposing I believe myself to be located at location q, I could entertain the thought 'here is it dark', and in this case my thought would be identification-dependent. The information link in this case, the feed through the video camera, is not one of the links that is constitutive of my 'here'-Idea. Thus my thought can be resolved into 'it is dark at q'; and 'q is here'.

Evans then points out that there are some thoughts, such as those involving Julius, that do not involve identifications, and so these two will be identification-independent. However, the interesting class of identification-free thoughts are those that are information-based. So 'identification-free' in the wide sense simply means 'not identification-dependent', and thus includes Julius-thoughts as well as 'here'- and 'this'-thoughts. 'Identification-free' in the narrow sense will be taken to mean 'identification-free and information-based', thus ruling out cases such as the Julius thoughts.

From the middle of p. 181 to the bottom of p. 181 Evans discusses a sense in which identification-free knowledge is basic. The argument here is quick. It is that if there is to be any singular knowledge at all, then not all thoughts can be identification-dependent. An identification-dependent thought, ‘a is F’, is one that can be analyzed as the result of two thoughts: ‘b is F’ and ‘a = b’. But if there are no identification-independent thoughts, then the first of these — ‘b is F’ — must itself rest on two thoughts, such as ‘c is F’ and ‘b = c’. And the infinite regress is off. At some point there must be knowledge that is identification-free if there is to be any singular knowledge at all. An argument similar to this, but given more detailed treatment, is provided by Shoemaker in the articles Evans cites in footnote 55.

The top paragraph on p. 182 discusses 'this'-thoughts and their identification-free status. Again, the point is that because 'this'-Idea involved in the subject's thought about this (a particular demonstratively identified object) constitutively involves a number of continuing information links, information from any of these links to the effect that property F is instantiated is shunted directly into the controlling conception of the demonstratively identified object with the result that the subject has information to the effect that this is F.

The second half of p. 182 to the bottom of p. 183 discusses 'here'-thoughts and their identification freedom. There are two sides to a 'here'-Idea: an informational aspect, and an 'action' aspect (see 6.3). The informational aspect, involving as it does the requirement that a subject have certain information links tied to his 'here'-Idea, work sin a way entirely parallel to 'this'-Ideas, with the difference, inessential for current purposes, that in the case of 'here'-Ideas the information links need not be continually delivering information. The action component, which is discussed at the top of p. 184, makes possible a kind of specification of actions that are free from identification errors. So for example, my intention to paint my car results in my decision to paint this car, and my belief that this car is my car. Nevertheless, there is specification of my action, my painting of this car, which is not liable to any sort of identification error (though of course it is liable to other sorts of error).

From the second paragraph of p. 184 to the bottom of p. 188, Evans discusses a potential objection to his claim that here thoughts are identification free. The objection is that since it seems that there are cases in which the information links do not automatically bring forth the corresponding judgment, such thoughts are not identification free after all. The example is the subject who has reason to believe that he may be wearing ultra-lightweight earphones, and so though he is getting information to the effect that ' ... is noisy' is instantiated, he does not automatically think 'here it is noisy'. If the subject then decides that the noisiness is not the result of the earphones, he might judge that 'here it is noisy', but then this might seem like a thought that can in fact be resolved into two thoughts: one of which involves attributing a property to some place, 'location q is noisy'; and another that involves an identity such as 'location q is here'.

Evans has a number of responses to this line of thought. The first involves putting pressure on the cogency of the supposed identity component. The 'location q' component seems only to be identifiable as the causal source of the information. This is fine (note that in the video monitor example, that is how we would identify the location being displayed on the monitor: it is the location that the video camera is trained on, and hence is the causal source of the information displayed on the monitor), but it cannot generally serve as supplying the content of our 'here'-thoughts. And it seems implausible to claim that all of our 'here'-thoughts involve this sort of conceptual equipment.

The second component of the identification, 'here', is an Idea that cannot be had if the only kinds of thoughts into which it can enter are identification dependent, that is, if there are no identification-free thoughts of the form 'here it is F', based only on information, gained in the appropriate way, that Fness is instantiated. Evans' argument here, I think, is as follows: If Evans' proposal that the content of 'here'-Ideas is involved with their provision of the possibility of identification-free thoughts, then the alternative must assume that this content is rather something like 'where I am'. But our 'I'-Ideas are bound up with our 'here'-Ideas in a way that renders this move impossible. The kind of identification-free thoughts that underpin 'here'-thoughts also underpin 'I'-thoughts.

Evans closes this discussion by pointing out that the case with the subject and the earphones shows that our normal 'here'-thoughts do rest on an assumption, but the assumption is not an identity assumption, but an assumption of the normal working of our perceptual systems. These system deliver information in such a way that it is shunted directly into the controlling conception of our 'here'-Ideas, whether we like it or not. We can question whether things here really are as they seem, but in so doing we are not questioning an identity assumption but questioning the functioning of our perceptual systems.

From the bottom of p. 188 to the end of the section, Evans discusses the relationship between his notion of identification free knowledge, and Shoemaker's notion of immunity to error through misidentification. The notions are closely related. For both Evans and Shoemaker, the core notion is that of a contrast between two kinds of thoughts: those that involve and identity judgment and those that do not. The difference is the criterion used to discern the two groups. In shoemakers case the criterion is this: A subject's judgment that 'a is F' is immune to error through misidentification if it does not make sense for the subject to wonder 'something is F, but is it a that is F?'. So for example, when I feel a toothache and think 'I have a toothache', it would seem to be senseless for me to wonder 'someone has a toothache, but is it I who have a toothache?'. whereas in the right circumstances it might seem perfectly reasonable for me to think 'I have a bump on my forehead' upon seeing a reflection in a mirror, and yet go on to wonder 'someone has a bump on his forehead, but is it I who has a bump on his forehead?', perhaps because I know that there are a lot of mirrors around at odd angles, and I may have been seeing someone else's reflection.

But Shoemaker's criterion will not work for Evans, for it will rule out some cases that Evans does not think should be ruled out. For example, the subject who believes he may in fact be fitted with the earphones might, upon hearing sounds that make it appear as though it is noisy where he is, think 'somewhere it is noisy, but is it here that it is noisy?'. More generally, Shoemaker's criterion rules out all cases in which one may raise a question about the correctness of the presumed applicability of the predicate of the judgment to the object (or location) that is the subject of the judgment. Evans, by contrast, wants only to rule out cases in which a question about the applicability of the predicate to the object of judgment is raised because there is an intervening identity claim that can be questioned. But he does not want to rule out cases in which the question can arise for other reasons, such as the suspicion that one's perceptual system is malfunctioning. So Evans suggests a rephrasing of the criterion:

First off, note that there is a significant typo in the text: The last sentence contains one too many 'not's. It should read: 'If the situation is perfectly normal, and the subject does not take it not to be normal, the utterance does not make sense.' Effectively this revised criterion modifies Shoemaker's criterion by adding a special background, that of an assumed normal functioning of the informational systems. Under such conditions, Evans takes it that a question can be raised concerning the correctness of the identification of the subject of the judgment only in cases where an identity claim is at work that might be false.

Appendix

Appendix Part 1

As McDowell notes, this material is drawn entirely from Evans' article 'Understanding Demonstratives'. Its purpose is to fill out a few discussions that were treated quickly in the text of Chapter 6. Specifically, those parts where the issue was either a) a subject's capacity to keep track of objects as they or the subject moves, either through space or time; or closely-relatedly b) the fact that demonstrative Ideas are temporally extended and involve a continuing information link. The point of the section is the notion of a dynamic Fregean thought, a kind of thought a grasp of which is a) graspable only via the orchestration of capacities that are manifested over time; and b) Fregean in that it is a particular way of thinking of an object that can be employed by more than one subject and be conventionalized so as to support a notion of Fregean sense, and also because Evans attributes the insight to Frege.

I will start with a sort of quick walk-through of this section of the appendix, and then move on to some remarks about the way in which it fits with various of the issues in Chapter 6.

Evans discusses Frege's remark that a subject can have, on day d₂, the same thought she had on the previous day d₁. On d₁ the subject thinks 'today is fine', and on d₂ thinks 'yesterday was fine', and thereby expresses the same thought. Evans will argue that although this can seem to be a slip towards inconsistency on Frege's part, it can also be seen as an initial step in the direction of an analysis of a very important and interesting conception of how thought relates to the world, especially for thoughts about the world that are dependent upon the subject's spatiotemporal location in the world.

Evans will argue that in the case above, the subject is thinking of the day in the same way in at least one relevant sense. First, however, he counters an argument to the contrary. The argument is that in such a case, it would seem that the intuitive criterion of difference could be used to individuate the thoughts as distinct, for surely it might be possible for a subject to affirm 'today is fine' on d₁, and deny or be agnostic about 'yesterday was fine' on d₁. Evans points out that strictly speaking this won't work, for the intuitive criterion of difference is defined only for a subject at a given time. It is not obvious how to extend it to the case of a given subject at two different times. Given that people change their minds about things, or remember or forget, it will be always be possible to find a difference of attitude at different times even in cases where there are clearly no new senses involved. (I might be agnostic on d₁ concerning the thought that Hesperus is Phosphorus, and accept it on d₂. Surely this fact cannot be combined with the intuitive criterion of difference to argue that two senses of 'Hesperus' or 'Phosporus' are in play.) Evans' specific argument is that in the case at hand, the criterion requires a prior decision on whether or not the same thought can be grasped at two different times, and the intuitive criterion by itself is no help here.

Next, Evans argues that a temporally punctate account of thought grasping is untenable. Such an account would claim that a sequence of thoughts that might be expressed as 'it will be F in a moment', 'it is F now', 'it was F a moment ago', are all completely different thoughts, and hence different beliefs (whereas on the account Evans is attributing to Frege, they are manifestation of the same underlying belief state). Evans argues that the ubiquitous succession from the thoughts of the first sort to thoughts of the second, and then the third, cannot be explained on the atomistic conception. Evans claims that the transition cannot be by inference. Presumably the inference would be something like 't₁ is F',(thought at t₁); therefore 't₁ was F' (thought at t₂). But in order for this inference to work, it must be possible for the subject to think of t₁ in the same way on at the two different times, otherwise the inference is invalid (a liability of any inference that employs ambiguous terms). But this ability is exactly what the atomistic theorist denies.

Next, Evans argues that the atomistic conception requires that the atoms are perfectly coherent, whereas the dynamic conception treats them as slices of the same persisting belief. The argument here is not clear, but it seem to be that any reasonable constraint on belief attribution would have to maintain certain minimal rationality requirements, and the putative subject entertaining the atomistic thoughts would not be making the minimal adjustments required to count as rational.

Evans turns to analogies with spatial thought, and how our ability to think of objects and places is dependent on our ability to keep track of the objects, places, and ourselves as any or all of them move about. Rhetorically the idea seems to be that these cases are much more intuitively clear, and the temporal case can be understood on analogy with the spatial cases. Evans' discussion here is clear enough.

Now to the connections between this material and that presented in Chapter 6 proper. Evans has maintained that Ideas of objects are underwritten by controlling conceptions, and in the case of demonstrative Ideas, these are information-based thoughts that are necessarily fed by a continuing information link. The fact that the Ideas give rise to thoughts that are identification independent means that they are such that the subject's receiving information that F is instantiated through the appropriate information link is sufficient for the subject to be disposed to think that the demonstratively identified object is F. But clearly this requires that the same Idea be temporally extended, for clearly as time progresses, provided we have not lost track of the object, we will continue to get, through the information link, identification-free information, and it is only if this information is fed into the same temporally extended Idea that it can be taken to concern the same object. Presumably the point here is that temporal indexicals are like 'here' in some relevant respects, they involve a special kind of Idea such that information links give rise to identification free knowledge (the perceptually arrived-at belief that 'now it is dark' cannot be felicitously seen at the result of two thoughts: 'there is some time at which it is dark', and 'that time is now'; in the normal case, it would make no sense to wonder 'at some time it is dark, but is it now that it is dark?').

The full picture, which Evans never got to articulate, is this: our indexical and demonstrative thoughts, including 'this', 'here', 'now', 'I', 'then', are underwritten by Ideas that have several features. First, they are associated with controlling conceptions that are constitutively tied to certain information links or (more generally) information channels. The controlling conceptions of these Ideas receive information via these channels over time in such a way that the Ideas involved must have a certain temporal thickness. This ought to be unobjectionable enough. Such thoughts concern the spatiotemporal world around us, and surely this requires the accommodation of temporal flow at the ground level. Second, these Ideas and their controlling conceptions have a certain kind of default tie to our spatiotemporal situation. In the spatial case, they are tied to (better, an aspect of) our capacity to keep track of where we are and how our relation to various places alters as we move about. In the case of objects, they are tied to our capacity to keep track of objects over time, as they and we move about. In the temporal case, they are tuned to track the normal passage of time (to within some degree of accuracy or other).

Appendix Part 2

This section concerns the role that sortals play in demonstrative identification of objects. It seems closely connected to the material in 6.5.

The topic is the extent to which sortal misidentification is consistent with having a coherent thought. Evans draws a distinction between two kinds of case of thinking 'that G is F'. In one case, one identifies the object (in some non-committal way) and is prepared to take it to be a G. In another sort of case, one identifies the object as a G. There is no real argument for this, other than the suggestion that it makes sense of certain kinds of case (which it must be admitted are themselves not clear). Specifically, Evans claims that one can have 'here'-thoughts in absence of any incoming information, provided one has the propensity to take any such information, should it arise, as relevant in the appropriate way to the 'here'-thought; but Evans denies that one can demonstratively identify 'this city' when one is shut up in a room. The cases appear to be parallel: one attempts an indexical/demonstrative thought about one's surrounding region when one is not getting any information from it. The only relevant difference would seem to be the sortals involved, and so Evans is under pressure to provide an account of the difference in terms of the relation to the sortals. But even here the projected argument is not clear, for the conclusion about identifying 'here' vs. 'this city' is one concerning whether one can do it at all; and the discussion of sortals here concerns when we would count misinformation as an error rather than an indication of the fact that the identification never occurred at all. That is, Evans would seem to need to claim that an attempted identification of something as a G in 'that G is F' requires some reasonable degree of information to the effect that it is in fact a G. If there is no such information, then one has no Idea and hence no thought at all. Hence the 'city' result. Since 'here' involves no special sortal requiring confirmatory information, 'here'-thoughts can be grasped in absence of information. The parallel account in the case of the woman would have to maintain that in the case where one thought 'that woman is F', and yet there is no information to the effect that the identified object is in fact a woman, then there is no thought at all — whether the entity really is a woman or not. Note that this is a stronger conclusion than the one articulated here in this appendix to the effect that the if the entity is in fact not a woman then the identification fails.

Appendix Part 3

In this section Evans discusses the phenomenon of immunity to error through misidentification in the case of information-based thoughts about objects whose mode of identification is descriptive.

Evans has previously argued for two claims. First, that the phenomenon of identification-free thoughts is a result of the way information-based thoughts work. In cases where an information link is such that its provision of information to the effect that property F is instantiated is directed directly into the controlling conception of some object, then the possibility of identification-independent thought about that object is provided for. Second, Evans has argued that there are cases of information-based thoughts whose mode of identification involves descriptive content. Here the fact that the word 'identification' is showing up in both claims is doing a disservice, for it means two different things. In the case of 'identification-free' it means that the thought in question cannot be discerned into two thoughts one of which expresses an identity. In the case of 'mode of identification' it is concerned with the resources the subject brings to bear in satisfying Russell's Principle. Evans argued in Chapter 6 that there are thoughts that are identification-free (in the identity-claim sense), and whose mode of identification (in the Russell's Principle sense) is demonstrative. But given that identification-freedom comes from being information-based and having appropriate kinds of information links, it should be possible for there to be identification-free thoughts whose mode of identification is other than demonstrative.

Take as an example of an information-based thought that exploits descriptive material in its satisfaction of Russell's Principle 'that phi was F.' In Chapter 5 the example involved seeing a big blue bird, and issuing an information-based thought about it later. The point here is that even though there is descriptive material showing up in the determination of the thought's content, the relevant aspect of the content of the thought is that it is information-based. Recall from Chapter 5 that this was the difference between 'the phi was F' and 'that phi was F'. Since we are assuming here (an assumption we need, but which Evans does not explicitly flag in this appendix) that the predicate is being ascribed on the basis of information received along an appropriately situated information link (and not on the basis of an inference from general principles, for example), the information to the effect that the object is F was mainlined into the controlling conception of the object, and was not put there on the strength of a belief that some otherwise-identified object is F together with a belief in an identity claim between the otherwise-identified object and the object identified as 'that phi'.

Given that the information to the effect that that phi was F came via a dedicated information link and not courtesy of an identification of that phi with some other object independently thought to be F, the identification-freedom of the thought would seem to follow immediately, and not require us to bring in considerations of intentions (of the sort described in Section 5.3) to refer to the object from which information derives. I am thus at a loss to explain why the argument in this section of the appendix takes the form it does.

Appendix Part 4

The topic of this section concerns demonstrative identification of abstract objects. Evans will argue that there are cases in which demonstrative identification of a token of a type can be a way of demonstratively identifying the type.

Here we get a very brief discussion of a topic that would require a good deal more. But it is suggestive nevertheless. The question is whether there can be anything like demonstrative identification of abstract entities. Evans focuses on types as abstract entities, and suggests that in some circumstances the demonstrative identification of a token of that type counts as a demonstrative identification of the type.

Evans begins by pointing out that on his account, one manages to think of an entity via the employment of an Idea of that entity. Such Ideas can be fundamental or non-fundamental. Evans' first move is somewhat questionable: he claims that the fundamental ground of difference of a type entity is the set of properties possessed by all tokens that instantiate that type. This seems questionable since, e.g., it would have us count spatiotemporal location as part of the fundamental ground of difference of colors, since all token-instances of colors have  spatiotemporal locations. (And Evans, in Chapter 4 said reasonably that the fundamental ground of difference of colors resides in their phenomenal properties.) In any case, this issue seems avoidable, for we can rephrase the criterion as: the fundamental ground of difference of a type entity will be that property that distinguishes it from all other types of the same superordinate type. For example, the fundamental ground of difference of red will be those phenomenal properties that distinguish it from all other colors. But since types don't, strictly speaking, have any phenomenal properties, this would be better stated in terms of tokens: the fundamental ground of difference of a type will be that property the possession of which by tokens instantiating that type distinguish them from the tokens of all other types of the same superordinate type. This phrasing captures what Evans needs, but without identifying all of the shared properties of the tokens of a type as part of the type's fundamental ground of difference.

Evans then suggests that a demonstrative thought of an abstract entity would be a thought employing an Idea of that entity that one has because one has in mind the fundamental ground of difference of that type in virtue of perceiving the relevant property instantiated in a token. This works typically by using an expression that designates a superordinate type, and using the token demonstrated to put one in mind of the property that distinguishes the particular subordinate type one is demonstratively identifying from the other subordinate types under that same superordinate type. So I might identify a color at a paint store by drawing the clerk's attention to a paint chip and saying 'Iwant that color'. (Notice the difference between a predicate used as a type noun, and used as a sortal. When pointing at an albino tiger: That animal is recovering from its illness ('animal' is sortal, demonstrative identification is of the token); That animal is on the endangered species list ('animal' is functioning as a type noun, the demonstrative identification is of the subordinate type of animal instantiated by the demonstrated token).)

Next, Evans tries to distinguish this sort of identification of a type from a descriptive identification, which would be something like 'the type of which this is a token'. The only consideration here is that there are cases in which perception of the token may be required for correctly understanding the remark, for having a fundamental Idea of the type. Telling the clerk that I want paint of a color that is the same color as the chip in my hand (when I do not show the clerk the chip), may allow the clerk to think true descriptive thoughts of that type. But exactly because it will not give him a fundamental Idea of the type, he will not understand the remark in the same way he might if I show the chip to him.

In closing let me point out that Evans might have been able to make his point more strongly. Many instances of such demonstrative identification of types do not involve identification of tokens that exhibit the relevant distinguishing property. In tutoring someone in mathematics, I might direct a student to some marks on a page and say 'this function, but not that one, converges'. I have put us both in mind of the correct mathematical function by identifying the appropriate symbols on the page — symbols which, of course, do not instantiate any relevant properties of the mathematical functions. Nevertheless, it remains true that I manage to get us to think of the relevant fundamental ground of difference of these functions via the symbols, and this seems to be the relevant factor. If this line of thought were continued, it might open up the possibility for Evans' account here to make fuller contact with Dummett's account of identification of abstract objects via a preferred set of symbols. Evans' account could explain what is right about Dummett's account in this case, but would be more powerful, because more general.

Appendix Part 5

This section rehearses five reasons that people find the account of demonstrative identification Evans has argued for in Chapter 6 (and previous chapters) difficult to accept. All five of these are discussed elsewhere in the text, but it is rhetorically helpful to have them discussed together.

The first is just the fact that some people do not distinguish demonstratives as a lexical category from demonstratives as a semantic category, and thus assume that all cases in which a lexical demonstrative, like 'this' or 'that' is used in English, the semantic category of demonstrative must be in play. Since such expressions are used in a great variety of cases, many of which resist analysis in terms of the apparatus Evans has developed, it may seem as though his analysis is unworkable. But if one distinguishes lexical from semantic categories, this problem disappears. Evans is giving a theory that recognizes and analyzes a certain semantic category. He need not assume that this semantic category is always and only marked by a lexical 'demonstrative expression'.

The second is a failure to distinguish what Evans calls 'situation-specifying' descriptions of mental states from 'content-giving' specifications. The hallucinator who thinks, as we might describe it, that the man in front of him is charging, can have his mental state described via a description of the sort of situation such that, were that situation actual, his mental state would have a genuine content. Such a  description is a good description of the hallucinator's state of mind. This is probably the biggest cause of resistance. The most convincing argument for Evans' position here is probably to be found in the Appendix of Chapter 5. But clearly a fuller defense of Evans' program here would be well advised to begin with this issue. For it seems quite natural to say that in describing the situation that 'approximates' the hallucinator's state of mind we are in fact giving the conditions that would render his mental state (to use a term not loaded as 'thought' is) true. And if this is correct, then it would seem that one is in fact giving the content of that mental episode.

The third is the failure to distinguish cases of misidentifying something that one successfully demonstratively identifies from cases where one fails to demonstratively identify anything at all. The distinction is one Evans has discussed in Chapter 6 as well as Part 2 of this Appendix. It is not clear why this is a source of resistance, however.

The fourth are cases where entities that do not exist are discussed using demonstrative expressions. We might talk about 'that spy' after seeing a movie. To some extent such cases will be discussed in Chapter 10, where an account of pretence is developed primarily to explain negative existential statements.

The fifth kind of resistance is a kind of adherence to a kind of representationalism that resists externalism of the sort embraced by Evans' position. This kind of representationalism holds that mental states, including thoughts, are entirely mental, internal states such that a) one knows whether or not one has that mental state, and b) whether or not one has it cannot be altered by changing external factors of which one is not aware. Evans' position denies both of these, for the existence or not of an appropriate object, whether the subject can tell the situations apart or not, determines whether or not the subject can have a demonstrative thought. Thus, I might take myself to be thinking a thought when I am not, and whether or not I am thinking this thought is partially determined by factors external (in some sense) to my mind. Evans suggests that something like the argument from illusion is responsible for the tight grip this kind of representationalism exercises. And there are some considerations here aimed at discrediting the argument from illusion. The argument from illusion is this: cases of illusion seem to the subject to be just like cases of veridical perception of a different state of affairs. Seeing a straight stick partially submerged in water looks like seeing a bent stick, for example. An explanation for this is that the subject has certain inner representational states, states proprietary to the subject's mind that determine the content of the subject's mental goings-on. Illusions are cases where external factors of one type cause representational states in a subject whose content would be specified via a situation of a different type. The partially submerged straight stick causes an inner representational state that, on its own, has a content involving a bent stick. Veridical perception on this model occurs when the external factors cause representational states whose content matches the external conditions in the relevant respects.

The relevant passage in which Evans responds to this line is here:

To be fair it can be said that this argument fails to join issue with the representationalist. In positing representations, the representationalist is not trying to double-count the fact that has already been accounted for — the fact that it seems to the subject as though he is confronting something. Rather, the representationalist is offering an explanation of this fact. The reason, the representationalist claims, that it seems to the subject as though he is confronting a bent stick when he is not is that a certain kind of representational state with a certain kind of content — a state normally or veridically caused by the perceptual presence of a bent stick — is had by the subject. Now this is not to say that other explanations of why it seems to a subject as though he is confronting something might not be offered, but no such alternate explanation is provided here.

Appendix Part 6

In this section Evans approaches the internalist position from a different path. Rather than discuss the similarities in mental states between a subject who is hallucinating and the same subject when not hallucinating, the issue now is the similarity between two subjects in different but analogous situations. The internalist argues that the similarity between my own thought that 'it is hot here' and the thought of someone else (located elsewhere) that 'it is hot here' is that they share a representational content. Evans argues that this cannot be right.

Evans begins by recapping one feature of his account of thinking of an object in a particular way. This account was that a subject S is thinking of x at time t in a particular way iff R(S, x, t). This is part of Evans' account of a theory of Fregean sense for demonstrative expressions. Another subject S' is then thinking of x in the same way as S if R(S, x, t).

But there is another sense of 'thinking of something in the same way' that can be articulated. It might be claimed that S' is thinking of x in the same way that S is thinking of x if R(S, x, t) and R(S', x, t). With the identification of this new kind of propositional function R(xi, zeta, t), it certainly is true that a given subject may not be able to tell the difference between his satisfaction of two different instances of this. For example if I (S) think 'it is warm here', thus satisfying R(S, x, t), I might be instantly transported to some other warm location x', and satisfy R(S, x', t). And I might not know that a change had occurred. Other kinds of change I might notice, of course, such as if I later think of x and think 'it was warm there' I am clearly thinking of the location in a different way, and in a way that I can recognize as different.

Given this, it might be suggested that a purely psychological aspect of 'away of thinking' can be extracted from this propositional function. This idea would be that R(S, x, t) could be analyzed into two components: a purely psychological component psi(S, t), and a non-psychological component R'(S, x, t). The notation here is not ideal, for designating this new relation R' suggests that it is a mere variant of the relation R, when in fact it is quite different: R (the relation figuring in Evans' preferred formulation) may include psychological factors — it might be a relation I have to x at t in part in virtue of paying attention to x at t, for example. R' by contrast, is merely a spatiotemporal context specification (or at least it need amount to no more than that), all psychological factors having been hived off into psi(S, t). This psychological component would be something I share with my doppelganger on twinearth, and would also have in situations of hallucination or illusion as well as the counterpart veridical situations. The proposal that this psychological component by itself counts as thinking. Evans will call this putative sort of thinking M-thinking. He will argue that this purely psychological M-thinking cannot really be thinking, indeed cannot even be contentful.

The argument is simple to state: a thought, or any representational state, is a contentful state. And having truth conditions is necessary for having content (the argument does not require the stronger claim that truth conditions are also sufficient for content). But the purely psychological predicates are not assessable as true or false. This is easily seen by noting that the same schema will apply in veridical and counterpart nonveridical situations as well as to myself and my doppelganger. The point hinges on singular content of demonstrative thoughts (and less obviously the singular content involved in almost all of our thinking, whether overtly demonstrative or not). My singular demonstrative thought to the effect that the coin I am looking at is valuable is, let us suppose, true. The psychological schema of this thought — the schema I share with my doppelganger on twinearth who happens to be looking at a quite valueless coin — has no truth value. It gets a truth value only when combined with the contextual element.

Evans then considers the following issue: historically such internalism was often motivated by a strong kind of mentalism, prototypically Cartesian dualism. On such a view, the mind, conceived as entirely distinct from the body and world, is the self-sufficient repository of the meaningful. But contemporary versions are motivated by completely different factors, including entirely materialistic factors having to do with the role of representations — understood as cognitive states with a purely neural instantiation base — play in the production and explanation of behavior. On this view, since it seems that me and my doppelganger, or me in the veridical and hallucinatory situation, will do identical things, we must have identical representational states. Evans points out that this argument works only by individuating the behavior that is caused or is to be explained in a question-beggingly individualistic way. Described in another way, the same behavior is not caused or in need of explanation at all: I seek a buyer for this valuable coin, whereas my doppelganger seeks a buyer for that valueless coin.

Evans closes with a rather cryptic argument that, I believe, more or less restates the essential point of the prior argument. Evans admits that part of the explanation for why I have the context-specific thoughts and intentions I have is that there are schematic dispositions that I share with my doppelganger. There is of course a kind of description of the brains of myself and my doppelganger that will yield the result that we are in identical states (of some sort), and these states will obviously play some role in determining the specific context-dependent thoughts and dispositions we have. But there is no argument for locating representational content or thought at this level as opposed to the more specified level.

 

[End of Guide to Chapter Six]

Copyright ©  Rick Grush 1999, 2000, 2001, 2002

Copyright holder grants permission to distribute, quote, and reproduce for any non-commercial purposes provided authorship and copyright are clearly indicated.