Philosophical Studies 39 (1981): 325-45. (Received 6 October, 1980)

MENTAL EVENTS

WILFRID SELLARS

I

1. I find that I am often construed as holding that mental events in the sense of thoughts, as contrasted with aches and pains, are linguistic events. This is a misunderstanding. What I have held is that the members of a certain class of linguistic events are thoughts. The misunderstanding is simply a case of illicit conversion, the move from 'All A isB' to 'All B is A'.

2. It is a familiar point that if we want to understand a certain kind of phenomenon, a wise strategy is to look for paradigm cases and get straight about them. Thus, if we want to understand what it is to be a physical object, we start from chairs, trees and pink ice cubes.

3. To be sure, it is equally essential to scrutinize borderline cases and contrastive items: shadows, echos, flashes and bangs. (Every dialectical enterprise has its negative moments.) And the ultimate aim is to have a clear grasp of the whole spectrum of the physical, from the middle-sized objects of the perceptual world to the strange entities of micro-physical theory.

4.Methodologically, however, one starts with what are, at least ostensibly, clear cut cases. In Aristotelian terminology we move from the better known to us to the better known in itself. For although the manifest world -- the Lebenswelt -- has its own intelligibility, it also has its mysteries. It poses questions which it does not have the resources to answer. The conception of the scientific enterprise as the search for the intelligible in itself is a theme which transcends the limits of the present occasion. I mention it simply to highlight the methodological point that in the domain of the mental, language it primary in the order of knowing.

5. In the domain of the physical, then, the middle-sized objects of the Manifest Image are prior in the order of knowing to microphysical processes. The latter, to the Scientific Realist, are prior in the order of being. In the domain of the mental, language is prior in the order of knowing. What, then, in this domain, is prior in the order of being?

6. There is no easy answer to this question, for I can think of no simple way of putting it which is not misleading. Perhaps the best way of putting it, of indicating the general character of the answer, is by saying bluntly, animal representational systems. For if the mental as linguistic is better known to us, it is in the theory of animal representational systems and, indeed, of representational systems generally that we come to grips with the better known in itself of the mental.

7. Like Chisholm, I take intentionality -- with a 't' -- a special case of intensionality -- with an 's' -- to be the mark of the mental. I agree with the classical view that there is a domain of 'inner episodes', properly referred to as 'thoughts', which are not linguistic -- though they are analogous in important respects, syntactic and semantic, to linguistic structures, and are functionally connected with linguistic behavior. These episodes have properties of the kind which are articulated by a sound phenomenology of mental events.

8. It should be clear that one can agree that there is such a domain, while leaving open the question as to its ultimate ontological status -- just as one can agree that there are numbers while leaving their ontological status undecided.

9. My disagreement with the classical view takes its point of departure from the fact that I construe concepts pertaining to the intentionality of thoughts as derivative from concepts pertaining to meaningful speech.

10. A language is, indeed, a means of communication. It serves as an instrument by which messages can be conveyed from speaker to hearer and our beliefs and intentions made available (or concealed from) our fellows. There are linguistic acts in the practical sense of actions. The philosophical study of linguistic actions launched by the late John Austin is flourishing.

11. But if there is a dimension in which linguistic activity serves as a means of communication, there is, as I see it, a more basic dimension, presupposed by communication in which it serves directly as a representational system, and is a medium in which we think. (In this context, of course, the concept of thinking is not the sophisticated concept of the classical theory, but the ur-concept which underlies the latter, as well as other theoretical concepts pertaining to various levels of representational activity.)

12. The central concept of this dimension of linguistic activity is, in my terminology, 'Thinking-out-loud' -- candid, spontaneous overt verbal behavior. Related concepts are those proximate, 'tip of the tongue' propensities to think-out-loud, and of other propensities and dispositions of varying degrees of remoteness (nestedness of conditionals) pertaining thereto.

13. I have developed and put to use the concept of thinking-out-loud on numerous occasions, and defended its fruitfulness against objections and misunderstandings. I shall not do so on the present occasion. Rather, I shall assume its intelligibility and exhibit its value as a means of clearing up puzzles about the mental.

14. Notice, once again, that I am not claiming that thoughts-out-loud and propensities pertaining thereto are what thinking primarily is in the order of being. I am saying, rather, that the concept of thinking-out-loud is our primary concept of thinking and is, therefore, our conceptual point of entry into the domain of the mental: as our concepts pertaining to the middle-sized objects of the perceptual world are our conceptual point of entry into the domain of the physical.

15. Our goal, of course, is to grasp what thinking is in the order of being. And to do this is, I argued above, to construct a general theory of representational systems.

16. Such a theory, which I propose to sketch, would enable us to understand how a language can contain a way of representing the world, i.e. how it can be, in certain respects a special and sophisticated case of a representational system.

II

17. Now, of course, animals, e.g. our useful friend the white rat, also have representational systems. They do not, however, have languages. Do they think? Have they mental events? Are they perhaps fraught with intentionality? Do they refer and characterize?

18. In order even to ask these questions, we must get clear about the meaning of the intentionalistic vocabulary of our methodologically primary domain. For if we go wrong here, our very questions will be based on a mistake.

19. Once again I shall have to make my points briefly and dogmatically. There is no time to do otherwise.

20. If our primary concept of thinking is that of candid 'languaging', it follows that our primary concepts pertaining to the intentionality of thought are syntactical and semantical concepts pertaining to language.

21. Now it has been thought that if this is indeed the case, then these concepts must be capable of definition in behavioral terms. Yet attempts to provide such definitions of 'meaning', 'sense', 'reference', 'denotation', -- not to mention 'truth', have been notoriously inept.

22. This failure has appeared to many to be a mortal blow to the very idea that there can be a methodologically primary sense in which languaging is thinking. For it would seem to entail that linguistic episodes have non-behavioral properties, thus those expressed by the above semantical terms, which are nevertheless essential to them.

23. Thus, the classical theory insists that the application of semantical concepts such as reference and aboutness in their various forms to verbal behavior presupposes that the latter is the manifestation of non-behavioral episodes of believing and intending which have an intrinsic intentionality which is intelligible in its own terms.

24. I have long argued1 that this is a mistake. The source of the mistake lies in the fact that semantical concepts have been taken to be relational, to formulate ways in which language is related to extra-linguistic reality.

25. Consider the shop-worn, but nevertheless fundamental, case of meaning. Ostensibly the contexts

E (in L) means...
E (in L) stands for - - -
are relational, i.e. contexts of the form
xRy
If they are, then the question naturally arises 'Is the relation in question definable in behavioral terms, or is it a unique non-behavioral relation?'

26. But the question does not arise, for its presupposition is false. It is easy to say that there is no meaning relation. The point is to understand why there seems to be a meaning relation.

27. To make a long and oft-repeated story short

'und' (in G) means and
has as its perspicuous regimentation
'und's (in G) are ·and·s
where the 'means' has been replaced by the copula and 'and' has been replaced by the dot-quoted expression '·and·s'. Meaning is not a relation because 'means' is specialized form of the copula, and, of course, the copula is not a relation word.

28. Why put the 'and' in quotes at all? because in the meaning statement it is obviously not playing the role of a connective and one peculiar way in which words play an unusual role is when, contrary to appearances, they are being mentioned. Why dot-quotes?2 because, obviously, German 'und's are not (ordinary quote) 'and's.

29. To think that quoting is simply a device for formulating illustrating names of sign-designs is to take far too myopic a view of the power of quotation marks. Quoting does form sortal predicates from linguistic exemplars. But the criteria for satisfying the sortal need not concern the design of the exemplar as such. They can obviously concern its function. Indeed the criteria can abstract from the design -- though they will require that items which satisfy the sortal have a design which is capable of embodying the function.

30. What is the criterion for being an ·and·? It is to be an item which functions in some language or other in a way which is relevantly similar to the ways in which 'and's function in our own, i.e. the background language.

31. Our meaning statement gives the meaning of 'und' (in German) by presenting us with an exemplar in our background language and telling us that if we understand how 'und's function in German we should rehearse in imagination the cluster of functions characteristic of 'and'. See, for example, the 'transformations' spelled out by De Morgan's formulae.

32. Now the problem posed by the ostensibly relational character of the context

E (in L) means...
arises in heightened form in connection with the context
E (in L) stands for---
thus
'dreieckig' (in G) stands for triangularity
'Schnee ist Weiss' (in G) stands for that snow is white.

33. Yet by an argument which begins by noting the strong equivalence which obtains between the latter statement and the meaning statement

'dreieckig' (in G) means triangular
'Schnee ist Weiss' (in G) means snow is white
it turns out that
to stand for triangularity is to be a ·triangular·
to stand for that snow is white is to be a ·snow is white·
'ity' and 'that' and a whole host of devices for forming abstract singular terms turn out to be ordinary language quoting devices which form ways of referring to items which play in some language or other the role played by the expressions to which they are added.
Triangularity = the ·triangular·
that snow is white = the ·snow is white·
and a reference to
the ·triangular·
is a general reference to ·triangular·s as a reference to
the lion
as in 'The lion is tawny', it a general reference to lions.

34. The gist of the matter is that 'triangularity' merely looks (to the eye bewitched by a certain picture) to be a name. It merely looks as though it refers to something non-linguistic: applying to expressions in any language which do a certain job, its inter-linguistic reference is confused with a non-linguistic reference. As in the case of 'means', 'stands for' merely seems to stand for a relation. It is, as 'means' proved to be, a specialized form of the copula.3

35. Now if all this is correct, then our primary concepts pertaining to intentionally can be shown not to concern unique modes of relationships between mental events and reality, but rather to provide a technique for classifying until events by reference to paradigms in our background language.4

36. If, therefore, there is a relevant degree of similarity between the functioning of a certain state, φ, of an animal's representational system and the function of 'this is triangular' in our own representational system, then we can appropriately say

φ-states mean this is triangular
φ-states stand for (the proposition) that this is triangular
i.e.
φ-states are ·this is triangular·s

III

37. But this functionalist theory of meaning and intentionality is but the prologue to a naturalistic philosophy of mind. It prepares the way for, but does not provide, a demystification of the place of mind in nature.

38. The crucial step is provided by a theory of predication which, taking its point of departure from the idea that for a predicate to stand for triangularity is for it to be a ·triangular·, i.e. to do the job done in our background language by 'triangular', tells us what that job is -- and does so in a way which gets us out of the familiar circle:

What is it to stand for triangularity?
It is to be a ·triangular·.
What is it to be a ·triangular·?
It is to do the job done in our language by 'triangular'.
What is the job done in our language by 'triangular'?
It is to stand for triangularity.

39. The theory of predication which rounds out the dialectic is, curiously enough, a form of nominalism which can lay proper claim to the 'universals are flatus vocis'. Of course, the role of predicates in a language is not to be pointless gusts of air. They have a 'meaning' -- let me say a function -- but that function is radically different from what it is usually taken to be.

40. Once again I must be brief and dogmatic. There is no other way in which I can make a distinctive contribution to this debate.

41. The standard view, of which there are many varieties, looks on the subject-predicate sentence 'a is red' in the PMese regimentation

red a
as a concatenation of two expressions 'red' and 'a', each of which has independent semantical status. 'a' denotes or refers to a; while 'red' refers to red things. According to the realist, the connection of 'red' with red thing is mediated by its connection with an abstract entity -- redness; according to the nominalist it is not so mediated. The contemporary nominalist is usually content to hold that the predicate 'red' is 'true of' or 'satisfied by' red things and that that is that.5

42. In general, the nominalist holds that there is no thing which 'red' stands for because it stands for red things and red things is not a thing.

43. Now it is a truism that the concatenation of 'red' with 'a' tells us that a is red. But is this fact illuminated by the idea that 'a' is correlated with a and 'red' with red things? I think not.

44. To understand what is going on, we need a different perspective on the syntactical form of 'red a'. I submit that it is properly viewed, not as a concatenation of two referring expressions, but rather as a name, 'a', which has the character of being concatenated to the left with a case of the sign design (flatus vocis) 'red'. If we abbreviation the expression

Concatenated to the left with a 'red'
by
red*
we can say with respect to a token, t, of 'red a' that while superficially
_
t is a 'red' 'a'
i.e. a 'red' concatenated with an 'a'; its true form, its depth grammar, so to speak, is given by
t is a red*['a'].
(Compare 'Fido is a white [dog]', where 'white' operates on the sortal 'dog' to generate a derivative sortal.)

45. Similarly, in the case of relational statements, for example 'a (is) before b'. Ostensibly we have before us something with the form

an 'a' concatenated with a 'before' which is concatenated with a 'b',
but its actual grammar is more perspicuously represented by
t is an 'a' and a 'b' with a case of the sign design 'before' (flatus vocis) between them;
or, abbreviating "with a 'before' between them" as 'before*',
t is an 'a' before* a 'b'.
(Compare 'x is a cat on a mat.')

46 Bluntly put

'φa' represents a as φ by virtue of 'a' being φ*,
'a R b' represents a and b as R by virtue of 'a' and 'b' being R*

47. This, we shall see, is the linguistic case of the more general principle

In a representational system, a symbol for an object, x, represents that object as φ by virtue of having a counterpart character φ*. Symbols for two objects, x and y, represent them as related in a certain manner by virtue of standing in a counterpart relation, R*.

48. Notice that in the linguistic case, the counterpart character φ* involves in subject predicate languages, the presence of an additional sign design (or flatus vocis). Thus, in our example φ* is the character of being concatenated to the left with the auxiliary symbol 'red'. But it could have been otherwise. We could say that a is red without the use of an auxiliary sign design -- for example by writing 'a' in boldface.

49. Again in our linguistic case the counterpart relation, R*, involved the presence of an additional sign design. Thus, in our example R* is the character of having a 'before' between them. But we could say that a is before b, by writing 'a' above 'b', thus,

a
b
This is not a trivial point, for the statement with the auxiliary symbol
a (is) before b
and the statement without the auxiliary symbol
a
b
have the same syntactical form, namely 'a' and 'b' in a counterpart relation.6

50. With this as a background let us take a look at what linguistic and non-linguistic representational systems have in common.

IV

51. An awareness of an item as green is a response (not necessarily overt) to the item as green. But it isn't an awareness of it as green, simply by virtue of being a response to it as green.

52. An iron filing can be said to respond to a green magnet as a magnet. It clearly doesn't respond to the magnet as green, for it would respond in the same way if the magnet were of any other color.

53. But though the filing responds to the magnet as a magnet, we don't say, except metaphorically, that the filing is aware of the magnet as a magnet. I shall refer to the sense of 'responds to as' in which the filing responds to the magnet as a magnet as the causal tense of the phrase.

54. By virtue of what is a response an awareness as? An interesting case of a response to something as something is that of a rat which has acquired the propensity to leap at panels with varieties of triangles painted on them. As in the case of the filings we can describe the rat correctly as responding to a triangle as a triangle. But we should be careful about concluding that simply by virtue of its training the rat has acquired an ur-concept of a triangle.

55. To suppose that it has reflects the common conviction that the connection between representational states and objects is a direct one-one correlation. Obviously, the representational state (the 'symbol') is correlated with what it represents -- but this correlation may essentially involve other correlations -- thus between it and other representational states and between representational states and actions.

56. Indeed, I propose to argue that to be a representational state, a state of an organism must be the manifestation of a system of dispositions and propensities by virtue of which the organism constructs maps of itself in its environment, and locates itself and its behavior on the map.

57. Such representational systems (RS) or cognitive map-makers, can be brought about by natural selection and transmitted genetically, as in the case of bees. Undoubtedly a primitive RS is also an innate endowment of human beings. The concept of innate abilities to be aware of something as something, and hence of pre-linguistic awarenesses is perfectly intelligible.

58. My point, then, is that a much greater degree of integration of responses to triangles as triangles into the rats RS is required before we can appropriately say that the rat has even the most primitive concept of a triangle. Primitive inferences would also be involved. Thus a rat's φ-state wouldn't be a state of representing something as a triangle, unless it had the propensity to move from the φstate to another state which counts as a primitive form of representing it as 3-sided or as having, say, pointed edges.

59. I have long insisted that inferential moves are an essential feature of representational activity. It is therefore important to realize that inference is at bottom, the sort of thing Hume had in mind when he speaks of the association of ideas.7 The two crucial points I want to make here are

  1. The representational states involved in primitive inference have propositional form.
  2. Propositional form is more primitive than logical form.

60. To have propositional form, a basic representational state must represent an object and represent it as of a certain character. In the case of sophisticated RSs we speak of a basic representational state as referring to an object and characterizing it as thus-and-so.

61. Given a map of, say, Michigan, it is quite proper to say that a certain symbol of a certain shape connected by lines with other symbols of other shapes picks out Ann Arbor, characterizes it as a large city and as related is certain ways to other large and small cities related by a nexus of roads.

62. But, of course, the map is a parasitical RS. It depends for its mappishness on its use by human RSs.

63. By virtue of what does a certain representational state, ρ, of, say, a white rat represent a certain location λ in its environment? Clearly by virtue of the fact that ρ belongs to a system of representational states ρi, ρj . . . so related that the system is structurally similar to the spatial structure of its environment, consisting of λi, λj.

64. But though structural similarity is necessary it is not a sufficient condition for mapping. The representational states must be so connected with each other and with the rat's locomotor activity that together they constitute what can not too metaphorically be characterized as a strategy for finding λ.

65. The root of the idea that symbol S represents object O is the idea that S belongs to a RS in which it is so connected with other features of the system (including actions) as to be the focal point of a strategy for finding O.

66. Or again, by virtue of what does a representational state, ρ, represent an object as being of a certain character, φ? The answer is, perhaps surprisingly similar. The organism has a strategy for finding φ objects. This strategy essentially involves inferential sequences of the form,

ψ here, φ nearby
e.g.
Smoke here, fire nearby.

67. Such inferential patterns are uniformities in the occurrence of representational states. Certain kinds of representational states tend to be followed (or to be followed by the absence of) certain other kinds of representational states.

68. In addition to inferential moves, there are representational analogous to perceptual awareness. Clearly a suitably trained RS can come to be in a 'This is a triangle' state by virtue of being irradiated by a triangular object.

69. I have also stressed that a RS can represent its own behavior. It cat represent itself as, for example, jumping. Under certain conditions the representation of itself as jumping acquires the propensity to trigger a jumping. In such cases the representation of itself as jumping becomes a primitive form of 'choosing to jump'. An adequate theory of RSs will discuss how the action-triggering valence is transmitted along a chain of representational states from 'goal states' to the representation of actions. [Compare the analogy of practical inferences.]

70. Another dimension is added by memory-retention. But given the rest this seems relatively unproblematic.

71. To sum up the points I have been making in this section:

To be an RS is to be a primitive or sophisticated form of a perceiving -- inferring -- remembering -- wanting -- acting organism. These features are essentially connected. Thus each of them is essentially involved in the referential and characterizing aspects of representational states.

V

72. I shall conclude this schematic treatment of RS by presenting in summary form the interconnections of the themes I have been discussing.

  1. The core concept of a mental event is that of a representational event.
  2. A basic representational event is an event which has two characters: one by virtue of which it represents an object in its environment (or itself); another by virtue of which it represents that object as being of a certain character.
  3. In the case of a subject predicate language, the referential character of the representational event involves a referential symbol and the characterizing feature of the event involves an additional symbol of which the role is typically misunderstood.
  4. The function of the referential symbol is to represent an object.
  5. The function of what might be referred to as the 'characterizing symbol', however, is not to represent something, e.g. φ objects, such that one is R to the other, but to give the representational event of which it is a constituent a character by virtue of which it represents its object as φ or as two objects one of which R to the other.
  6. The misunderstanding of the role of predicates as auxiliary symbols (flatus vocis) is the chief -- but not the only -- culprit responsible for two fundamental flaws in traditional conceptions of the mental.
    1. The notion that essential to mental events is the involvement of abstract entities.
    2. The notion that the presence or absence of a subject-predicate structure marks a radical difference between linguistic and non-linguistic RSs.
  7. It is the falsity of the latter notion that I have been concerned to emphasize in this paper. Its importance for the philosophy of representational systems, i.e. the philosophy of mind can scarcely be over-estimated.
  8. Putting it crudely, the fundamental thesis I have been advancing is that while pre-linguistic RSs do not have 'subjects' and 'predicates' they do share with subject-predicate RS the duality of the functions of referring and characterizing. The fact that in a subject-predicate language these functions involve separate subject symbols and predicate symbols is, from this standpoint superficial.
  9. All of the above is compatible with the idea that the presence in a RS of subjects and predicates makes possible degrees of sophistication which would otherwise be impossible. But to develop this point would require a discussion of logic-using representational systems.

VI

73. The subject-predicate distinction has often been thought to be the key to the distinction, not only between linguistic and prelinguistic RSs, but, more abstractly, between what have been called 'propositional' and 'pre-(or sub-) propositional' RSs. Thus the concept of a basic proposition is often identified with that of a representation which consists of a subject term and a predicate term.

74. But if the argument of the preceding sections is correct, nothing of significance hinges directly on the distinction between RSs which do and those which do not involve a distinction between subject and predicate terms.

75. Furthermore, we have seen that there is a legitimate sense in which an RS which doesn't involve subject and predicate terms can nevertheless contain propositions and intend states of affairs. For it can -- and, indeed, to be a RS at all, must -- involve representational states which perform both referential and characterizing functions.

76. A state which refers to a perceptual object and characterizes it as f, and which, therefore, can be classified functionally as a ·This is f·8 state, is the very paradigm of a propositional state, and while the sortal in terms of which we classify it is built from an expression in our background language, it must be remembered that such functional sortals apply not only to expressions in any language which play in that language a relevantly similar role to that played in our language by the dot-quoted expression, but, as we can now put it, to representational states in any RS which play in that RS the relevantly similar role in question.

77. This extended interpretation of the dot-quoting device was implicitly appealed to in paragraph 36 above, where it was said that a certain representational state of a trained rat could be said to be a ·This is a triangular· state, and hence to express the proposition that this is triangular.9

78. In order to carve nature at the joints, then, it would seem clear that the place where the cut is to be made is neither at the distinction between RSs which do and those which do not involve the subject-predicate distinction, nor between propositional and non-propositional RSs.

79. Where, then, does it lie? The answer is straightforward and should not surprise. The crucial distinction is between logic-using RSs and RSs which do not use logic, though their operations are described by mentioning logical operations.

80. Abstractly put, the distinction is clear. But in actual application the distinction is often blurred.

81. The logical operations in question are, in the first instance, those of propositional logic and, in second instance those of the quantificational logic which they make possible.

82. Thus we carve nature at the joint by distinguishing between those RSs which contain representational items which function as do logical connectives and quantifiers, i.e. which have logical expressions 'in their vocabularies', and those which do not.

83. If we call the former 'logical RSs' and the latter 'pre-logical RSs' we must bear in mind that there is a legitimate sense in which the latter contain propositions, involve the functions of referring and characterizing and, which is at least as important, the inferential role.

84. Let us take another look at the last of these points, for there is a widespread tendency to assimilate all inference to 'logical inference' and to assume that all representational systems in which inferences are drawn 'use logic'.

85. The point to be appreciated is that the inferences involved in a pre-logical RS have a form which is illustrated by the sequence

'Smoke here', 'fire nearby'
but not illustrated by
'If smoke anywhere, then fire nearby there'
'Smoke here'
'Fire nearby'

86. Abstractly put, the difference between

(ceteris paribus) whenever RS represents 'smoke here' it represents 'fire nearby'
and
(ceteris paribus) RS represents 'if smoke anywhere, then fire nearby there'
is obvious.

87. The former tells us, in traditional terms, that the RS "associates the presence of smoke with the presence of fire". Let me therefore call the sequence

'Smoke here', 'fire nearby'
a 'Humean' inference, and contrast it with the 'Aristotelian' inference which involves the quantified premise 'If smoke anywhere, then fire nearby there'.

88. A Humean RS can get from 'smoke here' to 'fire nearby' only by a route which involves base representational states and no logical vocabulary.

89. On the other hand an Aristotelian RS can get from the former to the latter either directly, by the Humean route, or indirectly, by what I shall call the syllogistic route, i.e. the route which involves, to adopt Quine's useful terminology, the standing representation 'If smoke anywhere, then fire nearby there'.10

90. Thus we can say, following Leibnitz,11 that a Humean RS which moves directly from 'Smoke here' to 'fire nearby' apes an Aristotelian RS which syllogizes. The former gets from its starting point to the conclusion it would have drawn if it had syllogized. As Leibnitz put it, "[animals have] a sort of consecutiveness which imitates reason".12 He emphasizes that this cornsecutiveness is also found at the human level.13

91. The case of negation is particularly instructive. Here the question is 'How can a pre-logical RS ape the use of negation?'

92. Suppose that the environment of a non-(logic using) RS contains only three shapes: triangle, circle and square. Suppose, also, that there is a kind of representational state which can be appropriately characterized as rejecting a basic representation -- as contrasted with failing to represent it.

93. It might look, at first sight, as though a RS which is able to reject a representation has a primitive form of negation 'in its vocabulary', i.e. is able to be in a primitive prepositional state of the kind 'not-p'. This, however, as we shall see, is to put the cart before the horse.

94. Suppose, now, that our Humean RS has the following propensity clusters: (ceteris paribus)

  1. if RS represents 'triangle there', it rejects 'circle there' and 'square there'.
  2. if RS represents 'circle there', it rejects 'square there' and 'triangle there'.
  3. if RS represents 'square there', it rejects 'triangle there' and 'circle there'.

The first of these propensities would be actualized by the concurrence of the following states

RS represents 'triangle there', RS rejects 'circle there', RS rejects 'square there'.
and similarly, mutates mutandis, in the other cases.

95. Suppose, now, an Aristotelian RS which does have a symbol for negation in its vocabulary For this to be the case, the RS in question must be such that (ceteris paribus) if it represents 'p' it rejects 'not-p', and if it represents 'not-p' it rejects 'p'.

96. Thus our concept 'RS has negation in its vocabulary' involves our concept 'RS rejects a certain representation' which involves our concept 'RS does not represent a certain representation'. But instead of our concept of RS as rejecting 'p' being equivalent to the idea that it represents 'not p', the latter . idea presupposes the former. The concept of rejection is more basic than the concept of negation.

97. Our Aristotelian RS also has propensities which correspond to the familiar De Morgan entailments and their quantificational counterparts. Thus it has the propensity not to represent 'not q' while representing 'p and [if p, then q)' and the propensity to reject 'p and not-q' if it represents 'if p, then q'.

98. Suppose, now, that our Aristotelian RS contains the following standing representations (for simplicity I use their translation into PMese):

(A') '(x) triangle at x .→. ~ circle at x . ~ square at x'
(B') '(x) circle at x .→. ~ square at x . ~ triangle at x'
(C') '(x) square at x .→. ~ triangle at x . ~ circle at x'

99. This RS can get from the state of representing 'triangle there' to the state of rejecting 'circle there' and rejecting 'square there', either by actualizing the Humean propensity cluster (A) (supposing it to have acquired it) or by actualizing the Aristotelian propensities illustrated by the sequence of representational states

(1)'triangle there'
(2)'(x) triangle at x → ~ circle at x . ~ square at x'
(3)'triangle there → ~ circle there . ~ square there' (2), Appl.
(4)'triangle there and (triangle there → ~ square there and ~ circle there)' (3), (1), C.I.
(5) '~ circle there and ~ square there' MP
(6) '~ circle there' (5), Simpl.
(7) '~ square there' (5), Simpl.

States (6) and (7) are, we may suppose, by virtue of the propensities involved in RS having negation in its vocabulary, accompanied by the rejection of the representations 'circle there' and 'square there'.

100. Thus we can say that the Humean, i.e. non-(logic using) RS has, given that it represents 'triangle there', the representational propensities with respect to 'circle there' and 'square there' it would have had, if it were an Aristotelian RS with the standing representation (A'). Once again we would have a Humean RS which 'apes reason'.

101. In saying this we mention the logical operations involved in syllogizing. But only by the confusion involved in what James called the Psychologist's Fallacy can we be taken to ascribe logical operations to the Humean RS.


NOTES

* Contributed to a symposium by that title at the annual meeting of the APA (Western), Detroit, April 26, 1980.

1 Most recently in: Naturalism and Ontology (Reseda, Cal., 1980). See also: Science and Metaphysics (London and New York, 1968), 'Language as Thought and as Communication'.

2 I introduced and explained the dot-quoting device in 'Abstract Entities', Review of Metaphysics (1963), pp. 229-269.

3 'Meaning as Functional Classification', Synthese 27 (1974).

4 For an explication of denotation in terms of meaning, see Science and Metaphysics, pp. 84-5.

5 For an exploration of this and related topics see Herbert Hochberg, 'Mapping, Meaning, and Metaphysics', and my reply in: Midwest Studies in Philosophy, Vol. II (1977).

6 See: Naturalism and Ontology, Ch. 3, for more on the non-triviality of this point.

7 For an account of inference as association, see my 'Action and Events', Nous 7 (19730.

8 For convenience in typing I shall, in what follows, use ordinary quotes in place of dot-quotes.

9 Clearly an account of how different tokens can express 'the' proposition that this is triangular -- in short a theory of demonstrative propositions -- is called for in order to resolve easily generated puzzles. Something like Strawson's distinction between a sentence and the family of statements it can be used to make on different occasions must be involved. But since 'proposition' has turned out to be a functional classification, this would simply amount to a distinction between a generic functional classification and a related family of more specific functional classifications. That it might not be possible to formulate the latter by the use of the dot-quoting technique serves only to remind us that the latter, however useful, is merely a useful dodge -- even though it has its counterpart in the ordinary language of intentionality.

10 In 'Some Reflections on Language Games', Philosophy of Science 21 (1954) [reprinted at Chapter 11 in: Science, Perception and Reality], I referred to such representations as 'auxiliary positions' and contrasted the moves in which they are involved with 'material moves'.

11 Monadology, §26. " Loc. cit.

12 Loc. cit.

13 Ibid., §28.