NotaBene е електронно списание за философски и политически науки. Повече за нас

What is „this"?

Rosen Lutskanov

Department "Logical Systems and Models"

Institute for the Study of Societies and Knowledge

Bulgarian Academy of Sciences



§1.My six year old son has lost something. Again. He is capable to mislay virtually everthing (yes, everything). As usual, soon after he has found that something is missing, he is ready to cry. He seems sure that he desperately needs the lost object, no matter what is it or whether he has even a vague idea where to find it. Nobody can reassure him with the philosophical remark that lost things usually reappear after a while. In practice, what this means is that I have to stop immediately what I am doing, no mater how important I think it is, and start searching. Soon I acquire the eerie look of a person who sees (what he is looking at) and does not see (what he is looking for) at the same time. We roam from room to room, opening lockers, picking up things at random and returning them back with frustration. When you can't find something, soon you get reduced to searching in really strange places, places where it is hard to put something, even in purpose, and even harder to search. That is why now I am laying on the floor, trying to reach under the sofa as far as I can get. Finally, I think I have sensed something with my fingertips and I proudly announce "I think there is something there", even though I hasten to add: "It seems that I cannot reach it". If I look, I shall see a cautious smile on the face of my son - the hope is back and the damned thing may be found after all. But he wants to know more about the mysterious entity I have just referred to. "What is this?", he asks in reply. Since I'm a philosophy graduate, I have (hopefully) mastered Quine's use / mention distinction, which means that I can render the question in two manifestly different ways:

  1. As "What is this?", i.e. as a question relating to the nature of the thing I was talking about.

  2. As "What is "this"?", i.e. as a question relating to the meaning of the word "this".

Nevertheless, I have never entertained even for a moment the second interpretation. As far as I can remember, my son has never asked me about the meaning of "this". As all other kids at his age, it seems that he got the meaning of this word without being supplied with a definition. This seems as one of the fascinating pieces of evidence in support of the thesis that some really smart people have put forth recently, the thesis that what we call "language" is a built-in capacity which is biologically hard-wired in all human beings[1]. How do you figure out the meaning of something so abstract without even asking for an explanation? I profess that I have no answer. What seems more important in the present setting is that the use / mention distinction seems to collapse with respect to "this": if you stick to the first interpetation, you should answer by providing a list of possible items, each identifiable by means of its shape and texture (for example, "I think it's a pen, a crayon or something like this"); if you switch to the second interpretation, you should answer by providing something similar - a list of different types of things which can be referred to (for example, "By "this" I mean a pen, a crayon or something like it"). The only important difference is that in the former context we loosely identify a (kind of) entity on the basis of some of its perceivable properties, while in the latter case we list different kinds of entities in order to elucidate the fact that the word "this" is a variable, standing for an indefinite object belonging to some definite category.

In a sense, we cannot really explain what is "this"; therefore we should be happy that our children do not ask for an explanation. If you don't know what "this" could refer to, then you would be incapable of engaing in or making sense of utterly simple language games like the particular game of searching that I have described above. Moreover, if you don't know what is "this", you could not refer, or point to things which is a basic cognitively relevant act. One of the first things (we think) we teach our kids is to refer to and interact with things and associate their appearance with specific words. When I say "Take this", I'm justified to suppose that I shall be understood properly, although this is a clear-cut case of radical interpretation of the kind that Quine has described[2]. The newborn is just like an anthropologist who comes to a strange tribe, the tribe of the grown-ups, and tries to master their language. Nevertheless, as far as we know, kids never identify rabbit colors, undetachable rabbit parts or rabbirt temporal stages as possible referents of the word "this rabbit". Their unfailing conceptual grip on words like "this" seems as a precondition for the development of their linguistic competence, if not as a precondition for thinking in general.

Of course, we are able to refer to other entities, not just things. We habitually substitute "this" for abstract objects, properties and relations, states of affairs and facts, events and occurrences. But it is a fact that children master first the application of "this" to concrete objects. The books for children are full of pictures of things, accompanied by their names. This is a sure sign that we, the grown-ups, think this somehow corresponds to the proper way of learning language. Maybe in different cultures, where children see not pictures of rabbits but real rabbits, who run away, scared by the sound of their footsteps, the things look different and events like "the running of the rabbit" have precedence to things like "the rabbit". There is no pressing need to resolve this issue at this point, since its answer does not affect what I am going to say about things. Things seem as basic in many ways - you can't pick up a property, describe a state of affairs, or relate an event to somebody who was not around when it happened, if you are not allowed to refer to objects or if you suspect that your interlocutor does not know what a thing is (It is completely different, and interesting, question what would be a reason to think that somebody does not know what a thing is).

§2.Up to now we have seen that even kids know what a thing is. Now we shall find that philosophers don't have even a clue about that. In its long and perplexing history metaphysics (or at least its most venerable department - ontology) was primarily interested in the definition of "thing". Take a concrete thing, i.e. a determinate entity with a plurality of inherent properties, say a red apple. How many things are referred to when somebody asserts that there is a red apple? Surprisingly, especially for non-philosophers, the classical answer is two - the particular apple and the universal redness (here it seems that the syntactical composition of the phrase "red apple" is taken as a sign for ontological structure). Plato and Aristotle seem to hold something like this, although they disagree which of those things are primary and which - secondary. For Plato, universals were primary existing things, hence particular things exist derivatively, by participating in their nature. The red apple exists insofar it participates in redness, applehood, and all other universals which mark one or another of its properties. For Aristotle, things were the other way around - in his view particulars were the primary existing things, hence universals exist derivatively, by being instantiated. The redness of the apple exists insofar it is instantiated by the red apple and all other red things. These two different stripes of classical realism were gradually driven out by the onslaught of nominalism, but another closely related doctrine was introduced relatively recently by Gustav Bergmann. The so-called substratum theory still isolates a pair of things in our red apple: a bare particular (the ingredient which makes this apple this very apple) and a property (the property of redness). But the bare particular is not a definite substance, it is a pure substratum, a peg, an empty placeholder to which different properties get attached, while properties are not exactly like Plato's otherworldly universals since they don't exist in detached form.

Other possible answers were explored as well. Another pair of opposed theories claim that the answer of the question above is "one". According to Russell's bundle theory, universals are the only real existent things and particulars are just bundles of compresent universals, i.e. universals which belong together in some way (Russell was really good in developing other people's theories to their logical end, or to the brink of insanity. Here he has outdone even Plato!). The opposed doctrine which appeared as a heresy but gradually reached the status of orthodoxy, is nominalism. It claims (roughly) that particulars are the only existent things and universals are constructions, classes of individuals, obtained by a relation of resemblance between individuals (this would be a succinct formulation of the so-called "resemblance nominalism", lots of other varieties of this doctrine were elaborated). As far as both the doctrine that individuals are unreal and the doctrine that individuals are the only existent things clashed with our (pre-)philosophical intuitions, a middle ground was sought. The uneasy compromise is called trope theory. Tropes are abstract particulars like "this apple's redness", they are abstract as far as they do not have independent existence and particular as far as they, unlike universals, are not predicable to many. So if Plato and Aristotle needed two types of entities (particulars and universals) and one relation between them (rendered as participation or instantiation), trope theorists postulated one type of primitive entity (tropes or abstract particulars) and two relations between them (the compresence relation of the bundle theorists and the similarity relation of the nominalists). For trope theorists, the whole history of ontology is an illustration of a confusion between two different distinctions: abstract/concrete and universal/particular, which prevented the realization that there could be abstract particulars, i.e. tropes[3].

As is well known, one person's common sense is other person's folly. Bearing this in mind I would readily admit that all these attempts to explicate the meaning of "this" are exceedingly far-fetched. Neither of these approaches with their countless modifications and elaborations comes close to the way I think of things, i.e. to the way I think most people think about them. In order to see how we think about things it seems utterly natural to explore the way we talk about them, since our language has the undeniable advantage to screen off otherwise insurmountable difficulties. Take some of your acquaintances and ask him: "I have just opened the refridgerator and there was just one red apple inside. How many things were there?". If he is not a philosopher, he would answer without hesitation: "One of course, you have just said so". If I insist: "Wasn't the universal "redness" also there?", the conversation would probably break at this point. This seems to disqualify the classical ontological approaches, along with substratum theory. Russell's bundle theory doesn't fare much better. The problem is that if bundles are sets of universals and sets are identified by means of their elements, then qualitatively identical but numerically distinct entities would be unthinkable. So the fate of bundle theory seems tightly interwoven with the principle of identity of indiscernibles - a metaphysical territory I'm not prepared to explore on this occasion. What about nominalism and trope theory? As for trope theory, it would imply that there is no definite answer to the previous question. When there is a red apple in the refridgerator what really is there is a vast multiplicity of tropes of the form "this-apple's-shape", "this-apple's-color", "this-apple's taste", etc. A pretty embarrassing conclusion. Now it should seem that nominalism is the only viable ontological position. Still, it is not very tempting after all, predominantly because of the peculiar way it treats universals. For nominalists universals are similarity relations. In the simplest case, they would be equivalence (i.e. reflexive, symmetric and transitive) relations - then "redness" would be nothing more than a way to split the collection of all (colored) particulars into two equivalence classes - the class of red things and the class of not-red things. Moreover, since (binary) relations are nothing but ordered pairs of elements, this would have the unpleasant consequence that if I eat the red apple from our previous example, this would somehow affect the universal "redness", which is identifiable with the class of red things. On this particular occasion I think that we should side with Plato by insisting that universals do not lead such a precarious existence, that they are definitely more stable entities than the nominalist would readily acknowledge.

§3.Up to now we have posed a paradigmatically philosophical question and then we have found that its traditional philosophical answers are defective in a number of ways. What is even worse is that they are manifestly incoherent with our linguistic intuitions. If the motive for this exercise was to install some healthy skepticism about philosophy's ability to solve most of its own vexing difficulties, it would have been trivial. We all know (or should have realized thus far) that the problems of philosophy (if there are such problems) are not solvable by purely philosophical means. But my aim is completely different: I shall try to fuse some important insights from the existing approaches and build on this basis a novel account which is more credible (at least in my own view)[4]. Just like trope theorists, I shall need a one-tier, flat universe whose inhabitants I shall call features. The English word "feature" derives from the Latin "factura" which implies that features are causally efficient entities, i.e. things that can be perceived, observed, or detected in some way or another. Moreover, I propose to treat actual situations as scenes, or finite collections of features, while the world as a whole would be either the collection of all scenes, or a grand scene, which encompasses all other scenes (there is no pressing need to choose one of these interpretations, although I would favor the first one). Now, as Plato has famously noted, we carve up the world by its joints and that is the way natural kinds are born. I would claim that the same thing happens on the level of individual objects: perceived scenes get carved up into pieces. Barwise has called the conceptual chisel we use in order to carve them up a scheme of individuation. This would be simply an equivalence relation on a scene, whose equivalence classes could be called items (an item would be simply a collection of features that come and go together and on the basis of this fact could be lumped together or treated as a whole). Situations change over time, therefore we shall need a way to represent time and change in the present setting. In order to make things simpler, let us augment all scenes with an additional "empty" feature 0. Then we can define scene morphisms as mappings between scenes f: A → B with the additional property that for any pair of features a and a' in A, a is in RA(a') iff f(a) is in RB(f(a')), where RA and RB are the schemes of individuation associated respectively with A and B and RA(a) is the equivalence class of a with respect to RA. What all this boils down to is the simple requirement that the features belonging to different items should not "switch" their places - the color of this apple cannot become the color of that apple even if they become qualitatively indistinguishable (remember that feaures are close relatives of tropes, so they are not shareable). In this case it would be easy to introduce identity morphisms, i.e. mappings of the form id: A → A such that id(a) = a for any feature a in A (these trivial transformations could represent on conceptual level the stability of scenes, the fact that they are able to persist in time). Moreover, if we assume that morphisms compose and that between a pair of scenes there is at most one morphism then every collection of scenes and scene morphisms becomes a category[5] (in the mathematical sense of this word). From now on I shall call such categories stages. Scene morphisms belonging to a particular stage could be used to mimic temporal precedence: the presence of morphism f: A → B would imply that the scene A is before scene B (since precendence is relative to a stage, we can easily incorporate in the picture relativity effects, distorting the time structure). If there is a pair of morphisms f: A → B and g: A → C but no morphism between B and C, this would mean that both B and C are after A but there is no information or no fact of the matter which determines the order of precendence between B and C (this may seem as a flaw but I prefer to keep the model incomplete in this respect). At this point our empty feature becomes useful: we could say that the item associated with RA(a) disappears in B if the image of the equivalence class associated with a is 0 (in the same way, an item would appear in B if the equivalence class associated with it is not an image of an item in A). This is in line with the interpretation of features that I've sketched above - an item appears if its features were not perceivable, observable or detectable before, it disappears if its features are no more perceivable, observable, or detectable (which is the characteristic mark of the empty feature). This should not come as a surprise: the thinking of changes of things requires the introduction of nothing as a specific kind of entity.

Now we have a descriptive language which provides resources for talking about scenes and items. This is just one of the things a real language should do, and real world objects are much more than our simple constructions that we have called "items". We are not simply individuating things (which is the job done by our schemes of individuation), we also classify them, which is an important part of our attempt to understand them and predict their future behavior. So alongside the schemes of individuation we deploy schemes of classification. They are collections of things that I call frames. A frame X is a collection of placeholders that I shall call slots (thus in important aspects it resembles a bare particular). Each slot has a domain of possible values which (as should be expected at this point) are features. Given a frame X and a scene A with scheme of individuation R an instantiation of X in A shall be a mapping v from some item x in A (i.e. an R-equivalence class) into X which is defined componentwise: for any feature a in x, v(a) is an element of the domain of some slot of X (it seems natural to assume that this mapping is an isomorphism, which would imply that the number of features of x is equal to the number of slots of X). Frames are types of items, and schemes of classification (collections of frames) are ways of seeing typological aspects of scenes (of course, a scheme of classification should be complete with respect to a given stage in the sense that every item in every scene of the stage should instantiate some of its frames, i.e. every item should belong to some specific type). Frames are my way to explicate the idea that individual things have an essence. In the language of frames and features this would mean that the features of every item have some limited range of variation which is identified by the domains of the slots they get mapped to by instantiation. This is simpler than it seems: if we conceive an apple as an item it would be reduced to a collection of different features (color, shape, size, etc.). But apples are much more than that, they have an essence. A red apple could have been green, which means that the slot for "color" of an apple has both "red" and "green" as admissible values. On the other hand, the domain of this slot does not include values as "black" or "purple", therefore no purple thing could qualify as an apple. Of course, we could paint an apple with any color we wish but this does not change what I'm trying to say here. On the one hand, when I describe something as a red apple this implies much about the tree that bore this fruit and about the other apples the tree is going to bear in the future. On the other hand, when I claim that a painted apple is purple I mean something different, which implies something about the causal past of this very apple and nothing more. Essences are more robust than simple descriptions since they are not restricted to actual features.

Furthermore, things are much more than items with essences, they also have a nature. This means that some of their features are related to some of their other features in some way or another. The color of a watermelon is a mark of its sweetness, the age of a person restricts the food she can eat, the brand of a wrist-watch determines at least to some extent its price. In other words, the features of a thing depend on each other. Dependence is a tricky concept, but I shall try to be more explicit on this point[6]. Take a frame X = {α, β, γ, ..., δ} and an arbitrary collection of items x = {a, b, c, ..., d}, x' = {a', b', c', ..., d'}, x'' = {a'', b'', c'', ..., d''}, etc. which instantiate X (I shall assume that the ith element of any item gets mapped to the ith element of X so that v(a) = α, v(b') = β, etc.). A typed identity is any statement of the form a≈a' which means that a and a' get mapped to one and the same element of the domain of α, D(α). A typed conjunction is a conjunction of typed equalities. Then δ shall be said to depend on α, β, γ iff the following holds: (x, x', x'', ...)(P  Q, where (i) P is a typed conjunction of equalities involving only the pre-images of α, β, γ in x, x', x'', ... (a pre-image of α in x, x', x'', ... is a feature of some of the items x, x', x'', ... which gets mapped to D(α) via instantiation, i.e. a, a', a'', etc. are different pre-images of α) and (ii) Q is a typed conjunction of equalities involving only the pre-images of δ in x, x', x'', ... . This general definition of dependency subsumes most of the better known kinds of dependency, for example functional dependency. It captures the intuition that δ depends on α, β, γ if you cannot vary the values for δ and at the same time keep the values for α, β, γ fixed. For example, the sweetness of a watermelon depends on its shape and color if you cannot find a pair of watermelons W and W' with one and the same shape and color such that W is sweet and W' is not sweet. I propose to explicate the nature of a thing by the set of dependencies it supports. In particular, take a scene A and an item x in A which instantiates a frame X. Take also the set C(x) of all scenes B with the property that there is a scene morphism f: A → B with the property that if x instantiates X in A, then f(x) instantiates X in B (these scenes would exemplify later stages of the temporal evolution of x in which it remains the same x, i.e. does not change its essence). Then the nature of x shall be encapsulated in a set of dependencies where x', x'', ... are images of x in scenes which belong to C(x). For example, take a scene A where there is an item s (Socrates) which exemplifies the frame H (human). Then if s has at A the attribute d (drinking hemlock), then there should be another scene B with morphism f: A → B where s is still H and f(s) has the attribute d (dead); moreover, there shall be a minimal B with this property, for which the following obtains: for any scene C, if there is a morphism g: B → C and s' = g(f(s)), then s' has the attribute d (if Socrates dies, he remains dead; this would be a case of persistent dependency, which is particularly important from the point of view of empirical knowledge). In other words, all these intricate looking conditions would mean that if a human being drinks poison, he would die, and if Socrates is conceived as a human being, then the constraint formulated above would apply to him (as we all know is the case). What this implies is that natures of things are determined by their essences, i.e. by the conceptual scheme we apply to classify them.

Finally, besides essences and natures, things have also a character[7]. The character of a thing is manifested in the systematic ways it interacts with other things which are present together with it. The machinery that we have elaborated thus far leaves room for an explication of character: for example, we can take the generalized form of dependency statements we have introduced above and generalize it even further by letting x, x', x'', etc. be different items instantiating different frames. If we substitute this modified form in the definition of nature, we shall obtain a definition of character, or a way of talking about the way the attributes of different items depend on each other. For example, if we light a match near the fuse of a bomb, the bomb would explode, which is indicative of the character of both matches and bombs. It is not a part of the nature of the bomb, taken as a thing in itself, to explode, since in the absence of variations of temperature and pressure in its immediate surroundings it would never do that. In the same way the explosion-provoking capacity of the match is not inherent feature of the match itself. This implies that the character of an item is relative to the stage at which it is located, since it determines the set of scenes to which it belongs and the temporal accessibility relations between them. The same is true of the nature of items since if stages are incomplete in some way (i.e. do not contain some scenes that they should), some dependencies may break down (remember the existential quantifier in the definition of "depends"). Even essences are relative, since they are defined by means of the instantiation of frames and nothing prevents a single item to instantiate different or even incommensurable frames. Socrates can be seen as an item instantiating both the frame "philosopher" and the frame "human being" but of course in the second case this would provide us with much poorer understanding of his nature and character, since the first frame is an elaboration of the second - it contains more slots and supports more dependencies. This exemplifies the simpler case when one and the same item is seen as instantiating commensurable frames standing in the species/genus relation. Much harder to treat is the case of toy-cars which instantiate a pair of partially overlapping frames - a toy car is a car (it has wheels, engine, etc.) but it is also a toy, i.e. not a real car (it cannot provide transportation, etc.).

§4.In a nutshell this is what my proposal boils down to: a thing, i.e. an entity which gets signified by "this" in the basic and primitive sense of this word, is an item to which are deployed schemes of individuation and classification, determining its essence, nature, and character (which in accordance to what we have seen above should be treated as irrevocably relative). What are the benefits of this particular articulation of the concept of "thing"? Here I shall try to list some of the most obvious advantages of the present approach by linking it to other areas in philosophy outside metaphysics[8]:

(4.1) First of all, it provides a more fine-tuned account of modality, since the phrase "the x might F" can be rendered in three different ways: (i) as related to essence - "The apple might have been green" means that the slot for color in the frame for apple has "green" as admissible value; (ii) as related to nature - "The ripe apple might rot" means that there is a dependency which links the slot values "ripe" and "rotting" in the temporal evolution of apples; (iii) as related to character - "The falling of the apple might provoke Newton's discovery of universal gravitation" means that there is a factual dependency between the features of different items (the ripe apple that has fallen from the tree and Newton's thinking about gravitational force and attraction).

(4.2) The present approach is focused on concrete objects - they are the "things" it tries to understand. This does not imply it is useless for the reconstruction of abstract objects. For example, what would be the universal "redness" from the point of view of the framed feature model? Well, we have considered different frames with a slot for color, whose domains may include different nuances of red - say "cherry", "rose", "crimson", "ruby", "scarlet", etc. Now imagine a meta-frame RED with just one slot whose domain covers all these different shades of red and a participation relation w, which links every color slot including some of these nuances with the meta-frame. Composing the instantiation mapping with participation mapping we shall get a relation between items (equivalence classes of features) and universals (meta-frames). Now an item x would participate in the universal "redness" iff it has a slot for color and the instantiation of this slot is mapped by participation to the universal RED. I think this would give at least partial justification of Plato's treatment of universals - a frame "exists" if it is instantiated, a meta-frame exists if there are frames with participate in it, without any relation to the question whether these frame are instantiated or not. In just the same way we can treat relations - they would be meta-frames with slots to which tuples of frame slots are mapped via participation. For example, the relation "higher than" would be a meta-frame which is linked to pairs of slots labelled "height" (a pair of heights would be mapped to "higher than" via participation if the value of the first feature is bigger than the value of the second feature] of course, this implies that features have values, conceived as participation mappings linking feature slots of frames with the abstract structure of numbers.).

(4.3) In the previous example we have touched upon another worrisome class of entities - the number. Just like other mathematical objects they are (justifiably) treated with some caution in contemporary ontology. We can iterate the approach that we've used above in relation to abstract objects like properties and relations and introduce numbers as a third type of entities linked to properties and relations by evaluation mappings. Thus we get a nice structure with four stages corresponding to different levels of abstractness: (a) stage 0: items which are built directly from features; (b) stage 1: natural kind terms (i.e frames); (c) stage 2: universals (meta-frames); (d) stage 3: mathematical entities, together with tree mappings: instantiation (linking stage 0 to stage 1), participation (linking stage 1 to stage 2) and evaluation (linking stage 2 to stage 3).

(4.4) Slight modifications can provide also tools for dealing with vague and incomplete objects. Remember that we've defined items as sets of features. This can be seen as characteristic for crispobjects. On the contrary, vague objects shall be composed of sets of features (this would necessitate the revision of the instantiation relation). Thus, if an object is vague in respect to color, it can have its color slot filled by a value like {green, blue} and there would be no fact of the matter which determines its "true" color. The treatment of incomplete (e.g. fictional entities) is much easier. In order to incorporate them we should introduce feature variables along with constant features. Consider for example the story of Ali Baba and the Fourty Thieves. Nothing in the story determines the exact eye color of the thieves, or their respective height and weight. Thus they would be items composed partly of constant features (corresponding to those of their determinations which are fixed in the story) and partly of feature variables (corresponding to those of their determinations which are not specified in the story). A second option would be to introduce contextual defaults - constant features which are temporarily introduced when the item is specified but are marked as subject to revision if necessary. Thus, having in mind that the background of the story is ancient Arabia, we are justified to introduce at the outset default values for skin color (say brownish), which are prone to be revised if some specific information is provided later in the story.

(4.5) We can treat states of affairs in the same way that we've treated things. They can be modelled as frames in whose slots are inserted not features but things (of course, they can be treated as meta-frames, just like properties and relations; this coheres with the rendering of states of affairs as abstract objects). If we include also the scene morphisms which stand for the way the things forming a state of affair evolve, we shall get something like meta-frame morphisms, which together with states of affairs form categories that I would call scenarios. Those meta-frame morphisms would capture one of the possible ways to treat events (as transformations of states of affairs) and so scenarios would serve as an explication of the key philosophical concept of process, i.e. a succession of events. At this level we could treat causality (which would necessitate a further generalization of the concept of dependence we have introduced in the previous paragraph).

(4.6) The framed feature model provides a natural way of treating inferential relations[9]. According to the subject-predicate model, propositions are two-element compounds. The relation of subject and predicate can be seen as a direct analogue of the relation between items and their features. Furthermore, the syllogistic model of inference studies inferential patterns relating a pair of propositions (premises) with a third proposition (conclusion). Why exactly a pair of propositions? The answer is readily available - because one of the premises and the conclusion state that the item (or collection of items) has some feature (or collection of features) and the other premise identifies the frame that those items instantiate. Turning back to the traditional example "All men are mortal. Socrates is a man. Therefore, Socrates is mortal" we can see what is really going on here. The subject matter of the syllogism is an item, identified by the name "Socrates". The major premise ("All men are mortal") identifies the nature of any item instantiating the frame "man" (i.e. it states that for any x instantiating the frame "man" in a scene A there is a minimal scene B and a scene morphism f such that f(x) has the feature "dead"). The minor premise states that Socrates instantiates the frame "man" and the conclusion builds on this fact in order to state that Socrates is mortal. This leads us to an asymmetric treatment of inferential roles of syllogistic premises which is justifed by the fact that the inferential transition from one feature to another could be sanctioned only by the nature or character of the thing under consideration. Thus validity of inference could be seen as stemming from those regularities that are seen as essential.



Fagin, R. (1982). Horn clauses and database dependencies. Journal of the ACM 29 (4), 952-985.

Koons, R. and T. Pickavance. (2015). Metaphysics. The Fundamentals. Oxford: Blackwell.

Lawvere. W. and S. Schanuel (1997). Conceptual Mathematics. A first introduction to categories. Cambridge: Cambridge University Press.

Minsky, M. 1986. The Society of Mind. New York: Simon and Schuster.

Pinker, S. (1994). The Language Instinct. New York: Harper Perennial Modern Classics.

Quine, W. v. O (1969). Ontological relativity. In: Ontological Relativity and Other Essays, pp. 26-68. New York Columbia University Press.

Люцканов, Р. 2015. Логика на асиметричната инференциална роля. В: Полименов, Т. и др. Логическата традиция. Университетско издателство „Св. Климент Охридски".

[1]Cf. for example (Pinker 1994).

[2]Here I refer to his well known remarks on the so-called "inscrutability of reference" (Quine 1969).

[3]For detailed overview of the multiplicity of current approaches, cf. Chapters 4 & 5 of (Koons and Pickavance 2015).

[4]My approach relies heavily on the way of thinking developed by Marvin Minsky. I've borrowed the concept of "frame" from his classic (Minsky 1986).

[5]For a good introduction to category theory (especially for non-mathematicians), cf. (Lawvere and Schanuel 1997).

[6]For more information on this issue, consult (Fagin 1982).

[7]Traditional metaphysics has largely neglected the topic of character. This is due to its excessive emphasis on natures, seen as inherent in things in themselves.

[8]Due to limitations of space, the list below is incomplete. There are ways to extend the current approach to cover different issues in logic and epistemology which shall be omitted here,

[9]I have treated this issue in greater detail elsewhere, cf. (Люцканов 2015).