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

Rick Grush (Department of Philosophy, UC San Diego)
Copyright  © Rick Grush 2003
Copyright holder grants permission to distribute, quote, and reproduce for any non-commercial purposes provided authorship and copyright are clearly indicated.

Chapter Eight: Recognition-based Identification

Table of Contents:

- 8.1 Introductory
- 8.2 Recognition and the Informational System
- 8.3 Recognitional Capacities and Space
- 8.4 Recognition and Recall
- 8.5 Recognition by Description
- 8.6 Mixed Ideas
- 8.A Appendix

8.1 Introductory

In this section Evans introduces the idea that an encounter with an object can be not only a source of information about the object, but can also allow the subject to gain a recognitional capacity for the object. Evans claims that in some (common) circumstances the possession of such a capacity is sufficient for a subjects having an adequate Idea of the object, and thus allows for the subject to satisfy Russell's Principle with respect to thought about that object. Evans brings up two objections to this proposal, addresses one in this section, and defers the other to 8.3.

Evans opens by discussing the issue of how one might think of something that one encountered previously, and from which one retains information. The first sort of case is where one exploits some individuating facts about the object to construct a definite description, including possibly the fact that it is the source of the retained information. But this is all prelude to the sort of case Evans means to focus on which is recognition: a case where one's previous encounter and the information one derived from that encounter provides one with a capacity to recognize that object if presented with it again. Having such a recognitional capacity for an object is sufficient, in some circumstances at least, for the subject to have an adequate Idea of the object:

In the context here it is clear that Evans could equally have continued "or whether or not the subject can currently locate the object in egocentric or objective space." That is, even in absence of any other means of discharging the requirements of Russell's Principle (e.g. description, demonstrative identification involving placement in space), an appropriate recognitional capacity can discharge the requirement under certain conditions, and thus provide or an adequate Idea of the object. Evans points out that there are cases in which recognitional capacities are clearly involved in our thought: he gives the example of observational concepts, such as red, where it is our capacity to recognize the color as red when presented with something red that is the foundation of our ability to think thoughts involving the concept red.

After pointing out that the kind of recognitional capacity he wants to make use of is not the same as a decidable description (this will be the topic of section 8.5), Evans turns to two challenges to the view that a recognitional capacity can provide for an adequate Idea and thus satisfy Russell's Principle.

The first, which he will mention now but not answer until 8.3, is that Russell's Principle requires discriminating knowledge in the sense that the subject must be able to distinguish the object of thought from all other objects. And it seems as though a recognitional capacity cannot do this, since it seems always possible that there might be more than one object that one would be inclined to 'identify' as the one previously encountered, a physical duplicate, for instance. Such a case would be one in which the recognitional capacity 'recognized' objects other than the right one (the one the previous encounter with which supplied the information that is the basis of the capacity).

The second sort of problem is the reverse: cases where the recognitional capacity fails to recognize the right object — for instance cases in which the object has undergone alterations of appearance that would foil a current recognition attempt. Evans begins with a parallel case: one's thought to the effect that some object was red. In this case, one's recognitional capacity for the property redness allows one to have an adequate Idea, so to speak of red, an Idea whose employment is not tied to thoughts about objects whose time indication is such that the subject can effect the recognitional capacity at that time. The point of the color example is to show that one's mastery of the color concept (the recognitional capacity) gives one the wherewithal to know what it means for something to be red regardless of the time index of the proposition. Thus, it gives me the wherewithal to entertain the thought that my office was red before I moved in, even though of course there is no possibility of me going back in time and applying my recognitional capacity to it at that time in the past.

Similarly, Evans introduces the notion of something's being recognizably a. Upon an encounter with an object a, one can gain a capacity to recognize a. This capacity endows the subject with a concept of something's being recognizably a, a concept which gives the subject an adequate Idea of the object, and which is applied to a again when recognized. If a goes out of existence, or changes appearance so that it would no longer be recognized by the subject as a, this does not obliterate the concept or the capacity to think of a by means of that Idea. One can entertain the thought [This was recognizably a], even though, of course, the subject would no longer be in a position to use her recognitional capacity alone to decide the truth of that thought.

8.2 Recognition and the Informational System

This section does not describe the workings of recognition and its role in thought so much as provide reasons why such a capacity has selective utility and is so important in our dealings with the world. In this section can also be discerned two distinct (but related) roles that recognition will play in Evans' project: the first is the ability to recognize objects and landmarks that set and define the cognitive map which undergirds our objective thought about the world. The second is the capacity to recognize specific people and places (and some other things that receive proper names) — this will play a significant role in Evans' account of proper names to be developed in Chapter 11.

After pointing out that much of our information about the world, both present and past, has been gathered in a way that makes essential use to recognitional capacities, either by ourselves or by those from whom we have received the information, Evans goes on to briefly discuss one particular application of recognitional capacities: the ability to recognize physical objects is arguably a prerequisite for a capacity to re-identify them, which itself is arguably a prerequisite for having a conception of an objective world. Evans cites Swinburne here, but he might as well have cited work of his own in Chapter 6, as well as his 'Things without the mind', and his teacher Strawson's work in Individuals and The Bounds of Sense. Evans does not expand on this issue here, which is unfortunate since it plays such a huge role in his program.

What he does expand on is the utility of recognition of other particular agents. His discussion opens with a consideration of the difference between two kinds of learning systems: a system that, on the basis of encounters with individuals, retains information only about the type of individual encountered. Evans' examples here are perfectly clear. The result of such learning would be that objects of type G are liable to perform certain kinds of acts with certain probabilities: dogs are very likely to bite; men with beards are quite unlikely to give me ice cream; and so forth. Contrast this with a learning system that, in addition to learning such propensities of kinds, is also able to initiate and maintain a dossier of information about individual agents. There are clearly many conditions under which being able to do so yields great utility. A subject might learn that while in general dogs are very friendly, the particular dog d is almost certain to be violent and bite.

While the utility of such informational dossiers initiated and maintained with a single particular in mind is clear, less appreciated is that in fact recognitional capacities equal to the task of reliably recognizing the individuals that are the topics of the dossiers is almost always a precondition for having such individual-targeted dossiers. After an initial encounter, when a dossier is started, all future encounters will need to be recognized as encounters with that same individual in order for the subject to be able to invoke the correct dossier — whether the  goal at the time is simply to add more information to the dossier, or to bring the information already in it to bear on the subject's thoughts and deliberations during the encounter.

At the top of p. 277, Evans lists some ways in which this system can break down or evolve: the dossier and the recognitional capacity may outlive the individual it is associated with, or outlive the reliability of the capacity's ability to recognize that individual; more than one object may, unknown to the subject, be 'recognized' as the same by the subject's recognitional capacity; and so forth.

Evans closes with a plausible explanation for why we are ready to allow recognitional capacities (and location in space) to trump descriptive modes of identification. Evans remarks in this final paragraph are sparse, but I think his point is this. Descriptive resources are built on the basis of conceptualizations of information that has been derived from one or more information channels. But it is the nature of many of these channels that misinformation is not infrequent: especially in the testimony system, but memory as well. In many such cases (though certainly not all) a conflict between an identification via spatial location or a recognitional capacity on the one hand, and descriptive elements on the other, signals a problem in the content of the information that is the basis of the description. Of course, it can work the other way as well in certain circumstances. This issue is one of the relative overall reliability of the accuracy of the information or the non-descriptive mode of identification, and Evans is here claiming that the latter have a significant edge most of the time. (E.g., if I meet someone and thus gain a recognitional capacity and also information to the effect that he is clean shaven, and a few weeks later I meet someone I am inclined to recognize as the same person but with a beard, my initial impulse is to trust the recognitional capacity and assume either that I misremembered the facial hair situation, or that it changed in the interim.)

 

8.3 Recognitional Capacities and Space

Here Evans does two things. First, he answers the objection he raised against his proposal in 8.1 but put off, to the effect that recognitional capacities cannot be sufficient to distinguish an object from all others because of the possibility of confounding duplicates. Second, Evans discusses the possibility of  'mixed Ideas' that have a recognitional component as well as some other component, such as descriptive or demonstrative. (This proposal will be modified considerably in the Appendix.)

Evans claimed in 8.1 that a subject's having a recognitional capacity for an object was sufficient for that subject's satisfaction of Russell's Principle — that is, it gives the subject an adequate Idea of the object. Evans there brought up the following possible objection: since we can never know that there are not exact physical duplicates of these objects, we can never be justified in claiming that these capacities allow us to distinguish the object form all others, as Russell's Principle requires. Thus recognitional capacities cannot serve as adequate Ideas of object.

Evans' reply to this is effectively the claim that there is more to a recognitional capacity than just that aspect that this objection focuses on. The objection focuses on what might be called narrowly recognitional aspects, based on the object's appearance. But Evans claims that spatiotemporal location is also an aspect of a recognitional capacity. (Evans claims, interestingly, that it is precisely the presence or absence of a location element to the recognitional capacity that distinguishes individual recognition from kind recognition, but this is not developed here.)

There are two kinds of location that can serve as the supplement to appearance-based recognition: egocentric and allocentric. The commonality is that in both cases, the location sets up a sort of default domain within which one takes it that one's narrow appearance-based capacity suffices to single out just one individual.

Evans' sheep example is of the egocentric variety. The radio example is of the allocentric variety. With these spatial considerations articulated, Evans cashes out (at the bottom of page 279 and the top of p. 280) what it is for a subject to have a recognitional capacity for some particular individual: i) the subject must be disposed to recognize x on the basis of its appearance; ii) there must not be another other individual y other than x in the relevant search domain that the subject is disposed to recognize with the same capacity; and iii) x is the right individual, that is the individual the encounter with which provided the subject with the recognitional capacity to begin with.

The addition of egocentric and allocentric spatial location criteria to recognitional capacities rules out the blanket objection to seeing recognitional capacities as capable of satisfying Russell's Principle. Of course such capacities can still be undermined by the unknown presence of confounding individual within the search domain, but this is a different sort of worry (to be addressed presently). And a requirement of this sort of recognitional capacity is the ability to make a number of practical inferences about the relevant search domain, based in large part on knowledge of how things, including oneself, moves in space.

So what if there is such a confounding individual? This brings up the second issue Evans discusses in this section. There are two sorts of case here. In the first, we imagine that the subject continues to assume that her recognitional capacity is adequate because she is unaware of the confounding individual. Though Evans is not entirely clear about this in the text here, the sort of case that is here envisioned in one where there is a doppelganger in the vicinity, but there have not been any encounters between the subject and the doppelganger. Cases where there have been encounters are discussed later on.

The second sort of case is one where the subject becomes aware of the fact that there is a confounding individual in the search domain. Evans is keen to argue that from the fact that in the first sort of case described above the subject would not have an adequate Idea, it does not follow that in the second sort of case the subject must lack an adequate Idea. The reason is that upon learning of the second confounding individual, the subject will, so to speak, switch Ideas from the one that she has just realized is not adequate to one that is. This new Idea is what Evans calls a mixed Idea. In this case, the mixed idea in question is based on the recognitional capacity along with a bit of conceptual supplementation of the form '... which I have met' or '... that I saw', or something of the sort that exploits the fact that the individual the subject means to think of is the one that was encountered. Evans is careful to point out that even if there are no other individuals, this conceptual material is available to the subject. But it is redundant, and in the normal case is not exploited in the construction or maintenance of the subject's Idea of the object. But because it is there, the subject has an immediately available fall-back Idea to exploit, a mixed Idea. (This will be discussed further in the Appendix.)

 

8.4 Recognition and Recall

In this section and the next, Evans tries to cement the central place of recognition in a theory of thought and reference by pointing out that it cannot be analyzed away as a kind of recall of a description. His point in this section is that even if recognition in some sense involves the use of information retained about an appearance (perhaps stored in sub-personal neural systems), it is not a memory that subjects typically can recall, nor does it appear to be the case that even if they could recall it, that any such image or description would be the basis upon which the recognition is effected. (The next section, 8.4, will argue that in fact there is reason to believe that recognition does not in any sense involve the use of a description, not even one stored sub-personally as a memory of an appearance.)

After some preliminary remarks, Evans begins with a distinction between two kinds of memory, recall and recognition. Evans describes the canonical psychological experiment demonstrating the difference: when presented with a list of words that one is to remember, there will be many items which, though one is not able to recall them later (if asked to write out a list of the words on the original list, the subject will not include these words), the subject will recognize them later as being words that were on the original list. This is part of common experience: if asked to name all of my colleagues at my former institution of employment, I am sure I would forget a few of the names; but if presented with a list including all of them and a large number of 'distractor' names, I would be able reliably to recognize all of them.

Evans then develops another example: our memory for routes and landmarks. My ability to recall the features of the route may be rather poor; but my recognition ability for items on the route, even crucial landmarks, might be very good.

Evans point with all this is not to quibble of the term 'remember', but is to point out that the information one can recall is often insufficient to serve to adequately describe or discriminate things that one in fact has a very good ability to recognize. My ability to recognize the right route to a location often requires much more by way of discriminatory power than I can recall, for example. Furthermore, the information that plays a role in the subject's recognitional capacities but which is not accessible to recall can not be part of what the subject is thinking. [Note: there is a typo at the end of the first paragraph on page 287. It reads "... with his having forgotten that there is such a route", but should say "... with his having forgotten that there is such a house." That 'house' is correct should be clear from the meaning of the text.]

With these points in hand, Evans claims that we must count recognitional capacities as providing for a capacity to entertain thought that cannot be understood on a descriptivist model. Evans provides an example of the man who thinks "That Russian was drunk." Given that the subject has a recognitional capacity for the object of his thought, we must accept that the subject really can entertain this thought. By the same token, the subject will be unable to recall features of the person sufficient to distinguish him from all others, as a description would have to do in order to be adequate to sustain singular thought. (Of course, the subject could very well think of the same person by description, as 'the man I saw last night in the corner drinking vodka', but Evans' example is meant to involve a subject essaying a recognition-based thought expressible felicitously as 'That Russian' as opposed to a descriptive thought of the form just mentioned. Evans' point is that the recognition-based 'That Russian' cannot be understood in descriptive terms — a description of the Russian's appearance; not that one could not think of what is in fact the same person by means of a description.)

At the top of page 288, Evans turns to his next point, which is that even if we could recall a detailed image or description, it would greatly falsify the facts of recognition to claim that we recognize something by recalling the vivid detailed image or description and comparing it to the object. The point can be made more strongly than Evans makes it. I can recognize any of hundreds of people as I walk through campus, where I typically see thousands of people walking about. Does it really make sense to think that for each of the thousands of faces I see I run through each of these hundreds of descriptions (the vast majority of such attempts yielding a 'no match' assessment), and when one of them corresponds to the description of, for example, the department chair I effect a recognition? Surely not. The mechanisms of recognition are best understood, as Evans claims, by neural mechanisms, not mental (e.g. formulating and applying a description) ones.

Evans closes this section by pointing out that his contention has been that even if the information that effects recognition can be captured in descriptive terms, it is not a description that the subject has access to, and so cannot be a part of the content of any thoughts that the subject might entertain on its basis. The reasons have been a) the subject simply typically cannot recall the information, and b) even if she could, it does not appear to be upon the basis of any such recalled information that we make recognitional identifications.

8.5 Recognition by Description

Having argued in the last section that recognition does not appear to involve anything like recall of an appearance, and that even if we could recall such information, we do not appear to make recognitions on the basis of recalled information, in this section Evans argues that the information contained in a recognitional capacity typically cannot be captured in descriptive terms.

Evans will argue that the information on the basis of which recognitions are effected cannot typically be captured in descriptive terms. Evans will use the example of the recognition of faces, and the first kind of proposal he canvasses is a crude one to the effect that we use descriptions such as 'receding hairline' and 'eyes widely spaced'. He points out that in fact it appears that these things are not the features to which we are responsive in recognizing faces. Rather, the nervous system appears to be sensitive to various ratios, such as the ratio of the distance between the eyes to the distance between the nose and the mouth. And again, this does not seem to be anything of the sort that subjects are conscious of when recognizing faces. The dialectic here is a bit odd, since Evans' point in this section was supposed to be that such descriptions could not capture the information, not that we didn't use such descriptions. The latter was the point of the last section. The point appears to be that if it can be captured in terms of a description, it does not seem to be the sort of description that most people would think of or be able to consciously apply. [Note: the neuropsychology of face recognition has called this specific 'ratio' proposal into question in the years since this was written, but Evans' point that the features to which the nervous system is sensitive are not open to the subject's introspection, and exploits sensitivity to features quite foreign to the subject's conceptual repertoire is unaffected.]

Evans then turns to the proposal that the description used is of the form 'x looks like this' or 'x looks more like this than anyone else I have met' where the 'this' is a sort of inner demonstrative aimed at an internally generated image of the face. In order to remove immediate reference to the subject's inner image, Evans' suggests that the proposal be re-phrased by i) capturing the relevant features of such an image in terms of its exhibition of some property j such that this property can be had by multiple things, and even exhibited by the internal; images generated by different subjects; and ii) replacing the private mental exemplar of this property j with an external publicly available image, such as a sketch.

Given this situation, can we make sense of the descriptivist proposal? Only, Evans claims, if a suitable 'looks like' relation can be secured which will allow such properties and exemplars to function as the descriptivist wants them to function. Evans points out that if there is such a relation, it will be secondary in Lockean sense: it will be 'x looks more like y than any other z to w'. Evans initially supposes that the w in this formulation is 'most people'.

Evans then turns to unpacking what is involved in such similarity assessments. Such assessments can be experimentally probed psychologically by probing similarity spaces. For example, we train an animal or person to produce some response upon stimulus A; and then we see which of stimuli B or C is better at eliciting the response. Evans suggests that this can be exploited to give an account of the relation between similarity and mental association, an account that turns the traditional doctrine on its head. The traditional doctrine, canonically stated by Hume, is that one of the factors that explains why the mind comes to associate D and E is that D is similar to E. Evans' proposal is that similarity can simply be defined as the capacity to make an association. Thus associations are primary. As Evans puts it (though footnote 37 qualifies this):

Next, Evans claims that an individual's reaction to something in such a way for her to think of it as similar to something else does not by itself constitute a judgment. By itself it is merely a reaction, and it obtains the status as a judgment only if it is, so to speak, taken up into the space of reasons, especially by being subjected to the assessment of others.

Given all of this, it seems as though there can be a perfectly legitimate property of looking like x that can be made use of in descriptions. The problem is that the proposal that a concept of this property is what people use in recognition faces the difficulty that there is no guarantee that the individual subject's reaction — whose recognitional capacity we are trying to explain after all — will be aligned with this public concept. As Evans puts it:

This absurdity means that we cannot regard any such public concept as being suitable for explicating a description that a subject uses to recognize someone. The recognitional capacity, if adequate, is immune to individual idiosyncrasy in a way that concept-application is not. Thus, the former cannot be just an implementation of the latter.

This invites (at the top pf p. 295) a retreat from the public concept to a private image and similarity reaction. One problem with this is that such a private concept cannot be made use of as part of a description capable of communicating contents, and thus is, as Evans puts it, useless to the description theorist. In addition, one can bring Wittgensteinian objections to this private concept.

The last paragraph sums the point of this section up nicely. What was wanted was some descriptive criterion that a subject could apply in order to determine whether or not some object or person was the object or person of his thought. Either this criterion is a public one with publicly answerable norms of application or it is not. If it is, then it cannot be of use for clearly a subject might recognize someone as being X and not Y even if that subject's similarity space is abnormal such that things that strike her as being like X strike all other people as being like Y. If it is not public, but private, then the proposal seems to get things the wrong way around. For in this case, it is not the fact that X has privately specifiable concept j that explains why S recognizes X; but rather it is the fact that S recognizes X that explains what it means for X to have the 'private concept' j.

It  might be wondered at this point what the relation is between this argument and Evans' earlier argument, in 8.1, to the effect that a recognitional capacity provides one with a concept of something as being 'recognizably a'. Recall, the point was to show that the recognitional capacity provides the subject with an adequate Idea of the object, one that can allow for thinking of the object even when the object is no longer recognizable by that capacity because it has changed appearance. While there may ultimately be a problem for Evans in making these two suggestions consistent, the clash is not as immediate as it might seem, for being sensitive to something's being recognizably a does not require that one have any kind of private concept of a property possessed by a in virtue of a's appearance. Rather, it need only require the appropriate sub-personal neural mechanisms that effect recognition. But the concept of being recognizably a is something that multiple people can grasp, though of course they may not all agree on its applicability in every case, and their grasp of it may be idiosyncratically tied to the sorts of neural mechanisms just mentioned. Another way of making the point is that the concept of being recognizably a is a perfectly good public concept, one that one can grasp in virtue of one's recognitional capacity for a (as 8.1 suggests) But it is not the fact that one is applying such a concept that one has a recognitional capacity for a (as 8.5 was keen to show) — indeed, as 8.5 argued, if anything it is exactly the other way around.

8.6 Mixed Ideas

In this brief final section, Evans expands on a topic first broached in 8.3: the possibility of a subject having an Idea of an object that is mixed in that it involves more than one way of knowing which object is in question. Such mixed Ideas are hybrids of more than one of the pure 'ways of knowing which'. Evans discusses cases which are, and which are not, decomposable into simpler Ideas.

Evans opens with an example of a thought that involves two ways of thinking of the same object: a judgment of identity between a man met on a previous day, and a man currently demonstratively identified. Such a thought might be expressed as 'This man is that man', where the 'this man' is a demonstrative identification of a perceptually present man, and the 'that man' is a memory-based thought, perhaps resting on a recognitional capacity. Upon making the identity judgment, the controlling conception of each of these Ideas will be merged: information gained during the previous encountered and in the dossier for the 'that man' object will be combined with the information in the dossier assigned to 'this man' on the strength of the demonstrative identification. This new controlling conception is the counterpart of a mixed Idea of the object.

Evans points out that in the sort of case just imagined the Idea, though mixed, will be decomposable into its components, because presumably the subject will be in a position to recover the original Ideas by segregating the information back into distinct controlling conceptions. This might be necessary if the subject learns that she has made a mistake, and that this man is in fact not that man. In such a case, Evans says, the subject is not in trouble, for she can recover her original Ideas in such a way as to make sense of the denial of the identity judgment. (She will have Ideas capable of serving on both sides of the identity operator.)

But it is also possible that the subject is not in a position to recover the original Ideas. For example, after a long period of time over which the subject had many encounters with what were in fact two distinct objects that she took to be identical. In each case, the information from the new demonstrative encounter was put into the same dossier, a dossier now containing information from two distinct objects, and such that the subject is unable to segregate the information into distinct files that correspond to the two objects. In this sort of case Evans claims that no coherent Idea of the object is available. Evans provides no special argument here, but his idea is surely the same as that exploited in Chapters Six and Seven, where he argued that in order to be well-grounded (and hence adequate) an information-based thought must identify as its object (via the mode of identification appropriate to that kind of Idea) the object that is in fact the source of the information. Ill-groundedness results when there is more than one causal source of the information associated with an Idea that is taken to single out a unique object.

Evans does allow, however, that in many cases it will be possible for a subject, when she learns about unknown duplicates, etc., to be able to recover at least some of the information into a dossier capable of supporting an adequate Idea: for example the information gained during the most recent encounter.

It should be noted that Evans discussion here is not entirely clear in one respect. He does not clearly distinguish between two ways in which information might be segregated. One way would be to separate them based on the kind of information-based thought that initially gave rise to the information. This sort of segregation might yield in one file all the information that was gained demonstratively from all the information that was gained via testimony. Another way to segregate it would be by the object of origin — as if one learned that on all the occasions where one was talking with a person one 'recognized' as Smith in the evening were with A, while all the conversations with the person 'recognized' as Smith in daytime were with B. All information may have been from similar kinds of information links, but the subject may simply be unable to separate the information out so as to be confident that the 'evening Smith' is the one who is married and has a summer home in Italy, etc., and the 'daytime Smith' is the one who likes to ski and has an allergy to pineapple, etc. While the discussion indicates that it is the second sort of information segregation whose success is crucial for re-establishment of adequate component Ideas of the objects involved, some of Evans' language inadvertently and misleadingly suggests it is the first.

It will also help to stay clear on the fact that there are two different kinds of problems discussed here that can be caused by surreptitious doppelgangers. The first is that any such doppelgangers in the vicinity can render the recognitional capacity inadequate, and render the Idea it supports inadequate (in Evans' technical sense). This first sort of problem is a problem even if the doppelgangers are not encountered. The second sort of problem can arise only if there are encounters with the doppelganger. It is that the information contained in the dossier associated with the object will end up containing information from more than one object, and hence thought-attempts based on the Idea will be Ill-grounded.

8.A Appendix

In the Appendix McDowell traces some lines of thought Evans probably would have explored more had there been time. These lines concern the right way to understand the aspect of recognition-based thoughts that renders them typically immune to the problem posed by duplicates elsewhere in the universe. In section 8.3, Evans explored the spatio-temporal search domain line. But here another line is explored, one involving the fact of the encounter between the subject and the object in the subject's past.

It is first pointed out that there is an analogy between memory-based thoughts (which include recognition-based thoughts) and demonstratives in that both are based on information, and also require a way of knowing which object is in question. Two ways that the second requirement fail to be fulfilled in the case of recognition-based thoughts are a) the subject cannot place the object in his own past, and b) the subject would not recognize the object if presented with it.

The body of the chapter did in fact maintain that having a recognitional capacity was sufficient for having an adequate Idea of an object (so long as this is understood to include the appropriate spatiotemporal considerations as described in 8.3). In the chapter it was maintained that under normal circumstances the subject would not resort to any sort of descriptive component of the sort 'which I have met' in order to supplement the recognitional capacity, but that such supplementation would be an available fall-back if the subject were to learn that there were a doppelganger in the vicinity. (Keep in mind that the sort of trouble case envisioned here is where there is a doppelganger, but the subject has not encountered the doppelganger. In such a case, the information is in fact from one object, but the recognitional capacity, on which the subject's satisfaction of Russell's Principle relies, will in fact not uniquely pick out one object. If the subject has unwittingly met this doppelganger and misidentified it, then an additional problem — ill-groundedness — is present in the subject's attempted thoughts.)

As it is put on p. 299:

Because of this, Evans/McDowell suggests that what was identified in 8.3 as the fallback position — the hybrid recognition and descriptive thought — is perhaps the normal case.

The large parenthetical remark that takes up the last half of the full paragraph on page 300 might mean something like this: there is a payoff to reconceiving recognitional capacities in this way. If we take recognitional capacities to have only a narrowly recognition component, and not to be mixed up with spatial considerations of the sort mentioned in 8.3, then we can appeal univocally to exactly these capacities in order to explain both singular recognition-based thoughts about an individual as well as recognition-based Ideas of natural kinds. An adequate Idea of a natural kind surely can rest on a recognitional capacity, but clearly need not involve anything like the idea of a spatiotemporal search domain based on an initial encounter as 8.3 suggests might be involved in recognitional capacities for particulars. With this narrower notion of a recognitional capacity in hand, we can now see the difference between the Ideas involved in natural kinds and those involved in recognizing particulars in this way: both involve a capacity for recognition (narrowly conceived), but the Ideas involved in particular recognition are mixed Ideas that also have something like a descriptive component of the sort mentioned in 8.3 as a fallback.

The final paragraph canvasses another potential advantage of this new conception. The argument of 8.4 and 8.5 centered around an attempt to preserve the Russellian status of recognition-based thoughts by arguing that they were not descriptive. One objection to this defense is that merely being non-descriptive is not a guarantee of being Russellian. For example, one might have a recognitional disposition for something that does not exist, in the sense that if that thing did exist, and the subject were presented with it, the subject would recognize that thing. But here it is claimed that on the new conception (on which a recognition-based thought aimed at a particular is a sort of recognition-descriptive hybrid invoking an encounter between the object and the subject), the Russellian status of these thoughts is secured.

It could be argued that this is perhaps not the best theoretical possibility here. It is true that these mixed Ideas will escape being hostage to the possibility of nearby unencountered doppelgangers, but it is not clear that it does so in a way that immediately secures their status as Russellian. For the non-existence of the object would not render the thought-attempt contentless, even on this new proposal. Indeed, a perfectly coherent notion of the content would be available: an object that would trigger the recognitional capacity and also such that the subject encountered it. Just as a recognitional capacity can be a disposition that would pick out a specific object if it existed, so too the descriptive component 'which I have met' would also be true of an object if I had met it. Adding these two non-Russellian ingredients together does not appear to result in a Russellian Idea.

But Evans is not without resources here, for it is not clear why the ideas developed in the Appendix to Chapter Five cannot have application in this case. There, recall, it was argued that the difference between 'the j' and 'that j' was a difference in the Ideas involved, but it was not claimed that the difference was to be cashed out by adding 'that I have seen' to the description. Exactly not. Rather, the suggestion was that the fact that the Idea is information-based is partially constitutive of the content of any thoughts made with the Idea's aid. The key passage is:

There it is claimed that the fact of the information link is partially constitutive of the Idea involved, but not by way of being descriptively added to the description j. If that suggestion is workable in the case of past-tense demonstratives, then it would seem to be workable here as well. One's memory- and recognition-based thought and employ an Idea that has its memory-borne information based-ness imprinted on it in such a way that only the object that is the source of the information could be the right object. This will make exactly as much sense, or as little, in this context as it did in the Appendix to Chapter Five. But assuming it is workable, it would seem to be the best way to understand how to secure the Russellian status of recognition-based particular thoughts in a way most faithful to the lines explored in the appendices to both Chapters Five and Eight.

[End of Guide to Chapter Eight]

Copyright  © Rick Grush 2003

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