Guide to Chapter Five

Guide to Chapter Five 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 5: Information, belief, and thought

Table of Contents:

- 5.1 Information-based thoughts: introductory
- 5.2 The information system
- 5.3 Interpretation and psychological attribution
- 5.4 The risk of ill-groundedness
- 5.5 Preview
- Appendix

5.1 Information-based thoughts: Introductory

In this section, Evans introduces the notion of a controlling conception of an object. In short, the controlling conception one has of an object is a bundle of content which bears on, and in some sense controls, the subject's thoughts about that object. An information-based thought is one which is governed by a conception of the object that includes information received from the object. A pure case is one in which the information derived from the current perception of the object (or one single current information source, to generalize away from perception to other information channels) exhausts the information in the controlling conception.

Remarks:

1. The controlling conception of an object may be quite extensive, perhaps of encyclopedic scope (as linguists would put it). It contains retained information about that object from past encounters, past testimony, information from current encounters and testimony, information that has been inferred, misinformation, and in general any information that is in a position to influence the subject's thinking about that object. The information need not be 'in mind' or be consciously reflected upon in order to be in the controlling conception.

2. A thought is information-based if the controlling conception of the object of the thought includes beliefs about the world which the subject has because he has received information deriving ultimately from the object. Given that perception, memory, and testimony are all parts of the information system, this means that most thoughts are information-based. Only in cases such as thoughts about Julius, or thoughts based entirely upon analysis (such as 'all natural numbers evenly divisible by four are even numbers') will be exceptions.

3. A 'pure case' will be one in which the information in the controlling conception is exhausted by information from the current information channel (current perceptual encounter, or one episode of testimony, etc.).

Examples:

A. When I identify someone I see as Otis, for instance, my governing conception of Otis will include not only information I gather from him when I see him on this occasion (he is wearing a red shirt, or whatever), but will include other beliefs held on the basis of past information-bestowing encounters, such as that he is a graduate student in philosophy, he is a good athlete, he likes beer. This conception will include, as a sort of bundle or dossier, whatever is relevant to my thought about Otis on any given occasion. These will be information-based thoughts because the controlling conception contains information which is a causal consequence of encounters with the object, either through perception, memory, or testimony. But this is not a pure case because there is more to the controlling conception than just information gained in the current perceptual information-gathering episode.

B. When I see a person that I do not recognize, such as a street performer, my thought about the person will be an information-based thought, but will be a pure case, since all the information in the controlling conception is being supplied by the current information link.

C. When you tell me about someone you met last night, the case is essentially the same as (B), but the information link is longer, having gone through your perceptual contact with the person, and then your testimony delivered to me. (Again, the case is similar to B assuming that I do not identify the person you are talking about as someone I have other information about.)

5.2 The informational system

In this section Evans discusses his notion of information at length (it is not the same as other more familiar notions). In the first part (up to about mid-page 124) Evans contrasts his notion of informational states with traditional sensation/belief epistemology. In the second part (from mid-124 to the bottom of 125) he discusses the content of informational states — the nature of the relation that holds between information and the objects from which the information derives. The final part (bottom of 125 to the end of the section) discusses features of the systems that deal with — gather, pass, and process — information.

There are three components to the human information system (the system that gathers and passes information) — perception, communication, and memory. Evans makes a point of describing the currency of these systems as information rather than belief. By way of contrast, Evans outlines a contrary proposal according to which what subjects do is to take as input sensations (something like the sense data of the classical empiricists), and then form beliefs about the objective world on their basis via some sort of inference. This was very much Russell's view, for example. Evans remarks that, contrary to this view, what is given to subjects of experience are seemings, experiential states already imbued with prima facie objective import, to the effect that the world is thus-and-so. Though there may be subpersonal states and mechanisms which operate with less than this, Evans claims that at the personal level, we work with seemings, and not mere sense data devoid of objective import. It is such prima facie objective seemings which are the contents passed by the information system.

Evans has a two-part argument for taking 'being in an informational state with such-and-such content' as primary, as opposed to 'believes such-and-such'.

First, it is often the case that we are subject to certain experience with some content, but we do not believe that content. He cites perceptual illusions (that we realize are illusions), and hearing fictional testimony (that we realize to be fictional). In both cases, it seems that we are in an informational state — a seeming that the world is a certain way — which we do not then go on to believe.

Second, one cannot try to analyze being in the non-believed informational states as mere prima facie inclinations to believe. The argument is that being in an informational state with such-and-such a content is something that we share with children and many animals. Belief is a far more sophisticated notion. We would have to credit children and animals with beliefs, and hence with reasoning, judging, and concept-applying capacities, in order to credit them with prima facie inclinations to believe, and it would be highly questionable to do so.

Evans then discusses the appropriate way to specify the content of informational states. His primary point is that in specifying the content of an informational state, one need not, and one should not, mention any specific objects. Rather, it should be specified by mentioning properties and object-variables (in an open sentence). This does not mean that we cannot perfectly well speak of the information being of this or that specific object. Two identical photographs might be such that one is of Bob and the other is of Rob, Bob's twin brother. But in specifying the content of the photograph, we should not mention Rob or Bob, but should rather say something like "a tall person with a red hat", a specification that both photographs would share.

Note that there seems to be a tension in what Evans is saying here. First, he claims that the content of informational states are seemings to the effect that the world is so-and-so. Given this, we must assume that informational states, even independent of belief, have something like objective import — they represent the world, the world represented as being independent of the subject, as being this way or that. But his likening informational states to photographic content, and his insistence that specifications of such content should not mention objects, but only object-variables and properties, seems to run counter to this. For surely photographs do not have a conception of an objective order. It might sound as though he is now saying that informational states' content is without prima facie objective import, because their content is not accurately captured by mention of genuine objects.

Probably the best way to make sense of this is to say that for Evans, the content of informational states is seemings that are poised to be objective. The reason that objects should not be mentioned is that mention of a specific object presupposes that the content includes identification information that distinguishes this object from all others (recall Russell's Principle and fundamental grounds of difference), and in fact informational states carry no such identification information. This is perhaps why, at the top of page 125, Evans says that the content can be specified by means of an open sentence in one or more variables (where each variable stands in for some, merely schematically specified, object), and not a mere Strawsonian feature-placing sentence (which would lack the variables). So the idea seems to be that an actual belief, (or more generally an objective thought), is the product of i) an informational state — a seeming; together with ii) the ability to objectify that seeming — to link the object variables in the informational state to Ideas employing fundamental identifications of objects in a way that satisfies Russell's Principle. It will be this second part that, e.g., animal cognitive systems lack. It might be wondered what the connection is between these remarks and the discussion of existential thoughts of the form some G is F from the previous chapter. Presumably the seemings that are part of the informational system are closely related to such existential thoughts involving only fundamental ideas d; and the jump to fully functional thought and belief involves something like the specific fundamental ideas d*. But the details are unclear.

Starting at the bottom of p. 125, Evans describes the minimal requirements for an informational system as being one that reliably produces states with a content that includes a predicative component. And he says that the informational system constituted by several intercommunicating agents can be such a system. Evans gives his diagram on the top of page 126, which has A perceiving something, retaining that information in memory, and telling B about it at a later time. B then retains this information in memory, and tells C at a still later time. Perhaps this was information to the effect that A saw a large blue bird. Evans then describes how, even without singular terms (singular terms require, apparently, the more sophisticated abilities that allow one to pass from mere information to genuine thought), information from different sources can be bundled and shuffled around:

a. an object is recognized, that is, an object one perceives is re-identified as one perceived at an earlier time.

b. an object perceived is re-identified as one that one heard about at an earlier time.

Evans then says that if we add singular terms to the communication system, then the next three mixed cases are also possible:

c. one now hears about an object, and re-identifies it as an object that one is now perceiving.

d. one now hears about an object, and re-identifies it as an object that one is currently remembering, from a previous perceptual encounter.

e. cases in which one now hears about an object, and re-identifies it as an object that one previously heard about.

This can be confusing, for why can one hook up two bits of information in the (b) case (where one hears about an object at t1, and then re-identifies it, at t2, as an object perceived at t2) without any singular terms, but in the (c)-(e) cases involving hearing about something at the later time, singular terms are needed? Perhaps part of what Evans seems to have in mind is that singular terms allow one to access an already-created controlling conception, or object dossier. Without singular terms, one can create a new dossier through the communication system, but not access an already-created dossier. For instance, someone tells you 'There is a big blue bird over there', you create a dossier for that object, and later when you see a big blue bird, you re-identify it as the one you heard about before.

But even this does not solve the problem, for if one can do such re-identifications in the (a) and (b) cases, then why would it not be possible to do it the other cases? Why couldn't you see a big blue bird, and then later when someone tells you that he killed a big blue bird, re-identify it as the one you previously saw (perhaps on the basis that you are convinced that this was the only big blue bird in the region)? In other words, why couldn't one access one's pre-existing dossiers even without singular terms? It is not at all clear why one could not get, via testimony, the sort of information that would allow one to access a dossier such that if got from perception would be sufficient to access a dossier. Provided that both systems give information that is specified by something like the open sentences Evans describes at the top of page 125 (for evidence that Evans thinks of the singular-terms-free testimony system this way, see the rest of section 5.2), and that one can make re-identifications on the basis of such content (as the (a) case exemplifies), it is not clear why the same could not be done via testimony, as in any of the (c)-(e) cases.

What Evans probably had in mind, but failed to express clearly, was a special way singular terms have of accessing object dossiers that is quite unlike, and in many ways more powerful and flexible, than the way they would be accessed by means of the content involved in mere open sentences. In all the sorts of cases I imagined above, what one does is to compare information one got from 2 information gathering episodes, and to re-identify if it seems as though there is sufficient match in the information. For example, I see a big blue bird, and later someone says something about a big blue bird. If there are very few such birds around, the match in content can be enough to access the previous dossier. But singular terms are linked to dossiers in a way that is independent of any content matches involved, and thus allow for identifications even when there is substantial mismatch or inconsistency between the contents provided from the various information sources. Looked at in this way, the real issue isn't that testimony per se is different, but that singular terms are special in terms of how they allow information to be accessed. The fact that Evans made two lists distinguished by the fact that the second has testimony showing up on the left hand side (c-e above) is not the clearest way this point might have been made.

But this is currently a side issue (it will take on central importance in Chapter 11), because Evans says he will concentrate on the pure cases, as in (a) and (b) for the time being.

Evans then argues that in fact the testimony system, if lacking singular terms, trades in existential contents like the photograph contents — that is, specifiable in open sentences in one or more variables. Such information might be information from (of) some particular object(s), but it would not result in a belief about that object. There is a nice parenthetical example on page 128:

(Consider my belief that swans are to be found in Uganda: is this about some particular birds, long since deceased, that occasioned the report in the travel book from which I derived the belief?) [p. 128]

That is, even though those birds were the source of the information, we would be hard pressed to say that the resultant belief was a belief about those birds. This distinction is of great importance for the rest of the book, for it is the distinction that i) undercuts the photograph model, which tends to conflate them, but ii) manages (in a way spelled out in the rest of this chapter) to capture what is right about the photograph model's insistence on the importance of information sources. (The fact that it is possible in some circumstances to have a genuine singular thought about those birds if one adds conceptual resources such as such as their role in the information transmission process (e.g. a definite description of the form 'the birds from which this information derives') does not affect the main point that need not be the case because of the nature of the informational content of testimony.)

Evans closes with some remarks on identity conditions for bits of information. He claims that match of content is neither necessary nor sufficient for identity of information. Rather, he seems here to be saying that a necessary (though he does not flat-out say sufficient) condition for identity of information is identity of causal source (this is the anti-descriptivist insight that adherents to the photograph model are right to embrace). He does say that 'we want to be able to say' that two informational states are identical 'provided that' they have the same causal source. This seems pretty close to a claim for sufficiency.

5.3 Interpretation and psychological attributions

This section is very straight-forward. Evans here wants to explain why in many cases, including the putative counter-examples to RP (the second steel ball, the polish grocer), we feel inclined to ascribe a thought about the object to the subject even though the subject has no discriminating knowledge of the object — an ascription that would constitute a violation of RP —while showing why such an ascription should not strictly speaking be made.

The case of the steel balls, for instance, was one in which though the subject did not have individuating knowledge — he could not distinguish the two balls in any way — many are inclined to say that he was thinking of the first ball, say, if it was the first ball from which his information derives. This appears to be in conflict with Russell's Principle, because we appear to be attributing a thought about one of the balls to a subject when that subject does not have the ability to distinguish that ball from all others.

Evans' answer to this is that in many cases, our use of psychological idioms, such as 'S is thinking about x', or 'S means x', is driven by our concern to make sense of the utterances of others by fitting them into what we take to be S's larger projects.

For example, the auto mechanics student who says, when questioned by the teacher, that the spark is produced in the carburettor. There is a clear sense in which the student's overall project and goals involve saying true things about the operation of the combustion engine, and given our knowledge of these goals, we find it natural to attribute to him the intention to say that the spark is produced in the cylinder. Saying this is what would satisfy his overarching intention. The other example is of the person who wants to register a complaint, and is told that he wants to see Mr. X. The person did not (in any strong de dicto sense) have an intention to see Mr. X. He did not know of Mr. X, and may not even have known that one needed to see anyone to register a complaint — perhaps he thought that a form needed to be filled out and thus didn't think any people at all would be involved. But it turns out that his intention to register a complaint requires him to see Mr. X.

The key point along the way is that it is not the case that when we ascribe to someone the purpose of referring to x, then we are ascribing to that person some thought such as 'I shall refer to x'. One can have an overarching goal or purpose which we as theorists can see would best be filled by referring to x even if the subject himself cannot see this, and cannot refer to x — as when the student intends to say 'cylinder' even though he cannot formulate such an intention. He may have never heard the word 'cylinder', and may have no idea what it is, having slept through the lecture on that day.

This is the first piece of Evans' response to the counter-examples. It is that we often find it natural to use psychological idioms such as 'S is thinking of a' or 'S means a' even in cases where the subject clearly can formulate, in thought or language, no such intention — provided that the subject has some clear higher-level project or goal or intention which is such that in order for that project/goal/intention to be successful, some reference to a would need to be made.

The second part of the response is Evans' claim that people have, as one of their higher-level goals, the goal of speaking or thinking about that object which is in fact the causal source of the information on which their beliefs are based.

With these two pieces in place, Evans' response to the counter-examples is clear. It is that yes, we do in fact find it very natural to say that the subject (in the example of 4.1) is thinking of the first ball when in fact the subject has no discriminating knowledge of this ball. But that is because the subject has, we can safely assume, the higher-level goal of thinking about whatever is in fact the causal source of his information. However, the subject is not in fact thinking of the first ball, any more than the student really intended to say 'cylinder' or the complainer had an intention to see Mr. X.

In connection with these points, Evans introduces the notion of well-groundedness: a thought is well-grounded if in fact it manages to concern the object that is in fact the source of the information in the subject's controlling conception of the thought. Being well-grounded thus involves a harmony between what the subject actually does (think about an object identified by some means or other, e.g. descriptive or perceptual), and what, given the subject's larger goals, the subject should be doing (think about the object that is the causal source of the information). This is parallel to saying that the student's answer is well grounded if what he actually says (that the spark is produced in the cylinder), is in conformity to his larger goal of saying something true.

5.4 The Risk of Ill-Groundedness

In the previous section, Evans claimed that it is an overarching intention of most speakers most of the time that their thoughts (and thus their linguistic productions) be well-grounded — that is, that they are about the objects which in fact are the causal sources of the information on which their beliefs are based. In this section he spells out a bit what he means by well- vs. ill-groundedness, in particular with respect to thoughts (as opposed to linguistic performances, which were the examples of 5.3).

Take for example an information-based thought whose mode of identification relies primarily on knowledge of individuating facts about the object (e.g. by description). An example might be That big blue bird that was down by the lake has surely flown away by now. In such a case, there are two factors to consider:

i) which object (if any) is the source of the information, and
ii) which object (if any) is uniquely identified by the descriptive content.

In the case of a thought whose content relies on descriptive material to pick out its object, the information-based thought will be well-grounded iff there is an object from which the information derives, and this object is uniquely identified by the mode of identification exploited by the descriptive content of the thought. Given this requirement, there are a number of ways that such a thought can fail to be well-grounded, and hence be ill-grounded. Evans enumerates some:

1. There is an object that is the source of the information, but no object is uniquely picked out by the content of the mode of identification. (There aren't, and weren't, any blue birds by the lake; it was a blue beaver mis-identified as a bird.)

2. There is an object that is the source of the information, and there is a unique object picked out by the content, but they are not the same object. (Like the case above, only where there coincidentally happens to have been a big blue bird by the lake, but the subject did not see it at all, but rather misidentified the beaver as a bird.)

3. There is no object that is the source of the information, but there is a unique object picked out by the content. (The subject didn't see anything — no bird, no beaver — but was merely hallucinating the bird. But there happens coincidentally to have been a big blue bird down by the lake at the time, unseen by the subject.)

4. There is no object that is the source of the information, nor is there an object uniquely identified by the content. (The subject was hallucinating, nothing was the source of the information, and there are and were no big blue birds down by the lake.)

The crucial cases are (2) and (3), for in these cases, there is in fact an object uniquely identified by the descriptive content, but that object is not the source of the information. Evans will argue that in such cases the thought is in fact not about the object uniquely identified by the content.

His argument is that in such cases, we would not say that the subject was thinking anything true in thinking that 'that j is F', when in fact there is a unique j, but it is not the source of the information — even when the unique j happens to be F. Everything in this argument hangs on the import of the demonstrative 'that'. The reason is that part of the content of the thought attempt is that the thought is an information-based thought (this is made clear in the appendix — see below), and this is the import of the 'that' in 'that j is F'. For example, suppose that the subject was down by the lake, and saw what was in fact a blue beaver, but misidentified it as a large blue bird. Coincidentally, there was a big blue bird at the lake at the time, but out of view. This bird was in fact foraging for food. The subject recalls the episode at the lake when he was looking at what was in fact the beaver and thinks 'that big blue bird was foraging for food'. The claim is that the subject will not have thought something true. Again, much hangs on the 'that', for Evans will be prepared to admit that if one were to think a purely descriptive thought of the form 'the big blue bird down by the lake yesterday was foraging' would be a true thought (where 'the' is an indicator of a definite description, and not another way of saying 'that'). For Evans, the 'that' is an indicator that the thought attempted is an information-based thought, and symptomatic of the overarching attempt of the subject to issue a thought about the entity, if any, which was the source of the information.

He goes on to argue that the conviction expressed above is not a manifestation of adherence to the photograph model. The photograph model, of course, would claim that the thought is about the object that is the casual source, and so would agree with Evans that the thought-attempt described above is not about the object that fits the descriptive content most closely, since in the case as described, that object is not the causal source.

Where Evans differs from the photograph model is on cases of type (1) and (2), where the PM, but not Evans, will conclude that the subject is managing to think something true. In the type (2) case, recall, we have a case where the subject misidentifies the beaver as a bird, but there happens, coincidentally, to be a big blue bird in the vicinity. In the type (1) case there is no bird in the vicinity, just the subject's misidentification of the beaver as a bird. According to the PM, the subject is successfully thinking about the beaver to the effect that it was foraging — the mismatch between the descriptive content of the subject's thought and the object that is the source of the information notwithstanding.

There are two parts to Evans' argument here. The first part is the point that, given the discussion in 5.3 concerning the use of psychological idioms, the fact that we would find it natural to describe the subject as 'thinking of the bird' or 'meaning' the bird does not entail that the subject really is thinking of the bird (any more than the ill-prepared student who has never head of a cylinder really meant to say cylinder, or was thinking of the cylinder).

The second point is that there is a problem in specifying the content of the thought that the PM would have us attribute to the subject. Let's change the example slightly. The subject is by the lake, and misidentifies as a big blue bird not blue beaver but a green blowing ball (some tricks of light filtering through the trees, and some ill-placed shadows, let's suppose). The subject later thinks 'that big blue bird was nearly spherical'. The PM would have us attribute to the subject a though, and true thought at that (the green bowling ball was, in fact, nearly spherical).

The diagnosis here is that as interpreters, we always realize, even if not explicitly, that subjects attempt to issue thoughts about the objects that are the causal source of their information-based thoughts. Most of the time this is successful, and there is a match between the descriptive content of the thoughts and those objects. In cases where there is minor descriptive mismatch, we tend to overlook this mismatch, and still attribute to the subject the thought about the object. Such cases lend intuitive support to the mistaken idea that being the course of information is sufficient to determine the object of thought. But when the mismatch between the descriptive content is more pronounced, it becomes apparent that the object is being 'thought about' or 'intended' in a way that is best accounted for via the interpretive mechanisms described in 5.3.

It is not clear how much mismatch, if any, is being allowed here. Much of the wording of this section seems to imply that any mismatch is enough to abort the thought attempt. In other places (especially footnote 21 on p. 135) the implication seems to be that a small degree of mismatch might be tolerable. It is not clear what the line is. One natural initial suggestion would center on Russell's Principle, and would take success at distinguishing the object from all others as the relevant criterion, and would work out from there, so to speak. So if the descriptive content employed by the subject would still pick out the correct object from all others even given the small mismatch, then perhaps a thought can be genuinely credited to the subject; and if not, then not. Even if this proposal is in line with what Evans would say, is still leaves unspecified what would count as 'still picking out the correct object'.

5.5 Preview

Evans here gives a brief preview of the rest of the book. He opens by pointing out that the argument of the previous section is a quite general defense of the Russellian status of any term an understanding of which requires an information-based thought, since if there is no object, the thought-attempt will be ill-grounded. Evans says that he will argue that many terms fall into this category: ordinary demonstratives, past-tense demonstratives, and what he calls 'testimony demonstratives'.

He will not go into descriptive identification (e.g. 'that big blue bird was foraging', where the mode of identification used is descriptive) in any detail in the book, because he says that we have a pretty good idea of what is involved (though there are some residual issues, as the Appendix will make clear). But demonstrative thoughts, and recognition-based thoughts are each given a chapter (6 and 8 respectively). Evans here says that there is a distinction between description-based identification on the one hand, and demonstrative and recognition-based identification on the other: in the absence of an object, the descriptive mode of identification still provides for an adequate Idea of the object; the descriptive content provides for an understanding of what would make the thought true. [Note that this last comment will be attenuated to a significant extent in the appendix.]

This sort of consideration for the Russellian status of expressions exploiting demonstrative and recognition-based modes of identification (where there is no object, there is no coherent Idea of the object) applies as well in the case of 'here' and 'I': and Evans will discuss these in Chapter 6 ('here') and 7 ('I'). So, all information-based thoughts are Russellian, since an information-based thought is ill-grounded if there is no object that is the causal source. In addition, the information-based thoughts of the demonstrative and recognition-based sorts are Russellian for an additional reason — in the absence of an object there is no coherent Idea of the object. Thoughts of these categories have their Russellian status over-determined.

Appendix

The appendix is more or less a re-working of the material in Section 5.4 in more detail, with some alterations. The upshot is that within all information-based thoughts, even information-based descriptive thoughts ('that j was F'), the fact that they are information-based is reflected in their content, and hence there is a difference in the content of 'that j was F' and 'the j was F', even though the descriptive component of the content is the same.

Evans points out that all information-based particular thoughts involve a 'duality of factors'.

1. Information, derived from the object, that the subject takes as germane to the evaluation of the truth of the thought.

2. The subject's ability to identify the object that his thought concerns, in a way that allows for satisfaction of Russell's Principle.

In the case of past-tense demonstratives ('That j was F'), the 'that' has the import of making it clear that this thought is meant to be based on information from the object, while the 'j' is the content used by the subject to identify the object in his thought.

Even though the subject identifies the object by means of (in this case) some descriptive content, the overriding purpose of the subject is to identify the object from which the information derives. This is a key point. It is the subject's 'overriding purpose' to speak about the object from which the information derives in the same sense that it was the student's overriding purpose to say true things about the combustion engine. The student did not need to reflect on his situation and think 'I wish to say true things', rather, the saying of true things is part of his unreflective practical comportment. The same is true for the subject attempting an information-based thought: he need not and often will not either i) think anything like 'I wish to think about the object from which my information derives'; nor ii) need the subject add a description of this information-link to the descriptive content of his thought. Though the subject is using the descriptive content to identify the object in thought, the subject is operating on the inexplicit practical assumption that the object so identified is in fact the object from which the information derives. This means that this goal can be attributed to the subject (in the sense discussed in section 5.3), even though the subject may not be using anything like a description such as 'the object from which this information derives' as part of the content he is using to identify the object. (By contrast, in the case of descriptive names like Julius, or even pure definite descriptions not associated with a name, such as 'the tallest person in California', there is no sense in which the subject has the intent of aiming at any object other than the one picked out by the descriptive content.)

If we call the object that is the source of the information the target, and the object that is singled out by the content the object, we can say that an information-based particular thought is well grounded only when the target is the object.

In a key passage Evans says:

The content is determined by the subject's mode of identification (the second factor in our duality); the fact that the thought-episode is causally related to the object in question (the first factor) serves, not as a determinant of content, but as a necessary condition for the possibility of ascribing an information-based particular-thought with the content in question to the subject. In giving the content of the thought, we need to specify the object in a way which mirrors the mode of identification which the subject employs; hence, not as the object which bears the appropriate causal relations to the information on which the thought is based. [p.139]

The claim here is that the content of the particular thought 'that j was F' is provided by the descriptive content j. The causal/informational factor does not contribute to the content, but is a precondition for there being a thought of that form at all (specifically, that object being the object picked out by the descriptive content is the precondition for the possibility of that thought). But that raises a problem, which is brought up in the following passage:

Surely the difference between the thought that the j was F and the thought that that j was F is a difference of content, and not a difference which can be wholly shunted off into some sphere of considerations external to the determination of content. [p.140]

But if the proposal expressed in the first quoted passage is correct, then the two thoughts ('the j was F' and 'that j was F') do have exactly the same content, that supplied by j. Evans' attempt to address this problem is perhaps not entirely satisfactory. This is on the paragraph bridging p. 140-1, and the first full paragraph on p. 141. The proposal is that we need to 'make sense' of the notion of a 'mode of identification that is, as it were, all but exhausted by a definite description.' This seems fine, but one wants to know what this surplus is, since as has been pointed out, it cannot be a description or concept of the causal information-link itself. The surplus, apparently, is not to be explained in terms of additional descriptive content, but rather in terms of the Idea employed in the thought. It is not clear what to make of this either, and we get no help from the text. Perhaps one way to understand it is as follows: one aspect of the Idea employed in information-based thoughts is that such thoughts are amenable to a certain kind of practical requirement that Ideas used in other thoughts (such as purely descriptive) are not: specifically, they are such as to be subject to the norm of being well-grounded. This is a norm that such Ideas are subject to not because of any additional descriptive content, but because of practical norms governing the way in which such thoughts are acquired and assessed.

In the penultimate paragraph (bridging pages 141 and 142) Evans produces an argument against assimilating this surplus content to the descriptive content (by saying that it is in effect adding '... and from which my memory is derived' to the description). The argument is that if this were done, then two subjects could not share a thought about the object, since each would be employing a different description, and hence a different content. But if the difference can be made out in terms if a special kind of Idea, then this might be a kind of Idea that more than one subject can use.

[End of Guide to Chapter Five]

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