Navigazione – Mappa del sito

HomeNumeri76The Dog Schema

The Dog Schema

Umberto Eco, Maurizio Ferraris e Diego Marconi
Traduzione di Sarah De Sanctis
p. 10-39


The Dog Schema is the translation of the Italian text Lo schema del cane, originally published in Rivista di Estetica in 1998. The text presents the debate among the philosophers Umberto Eco, Maurizio Ferraris, and Diego Marconi concerning the dog schema. The debate develops through a reflection on Kantian schematism and conceptual applications, showing three different points of view on the subject.

Torna su

Note della redazione

This text was originally published in Italian with the title Lo schema del cane in “Rivista di estetica”, n. s., n. 8 (2/1998), XXXVIII, pp. 3-27. The editors would like to thank Professor Diego Marconi, Umberto Eco’s family and the publisher Rosenberg & Sellier for granting its publication. The texts of Maurizio Ferraris and Umberto Eco have been translated into English by Sarah De Sanctis.

Testo integrale

Maurizio Ferraris - Introduction

1Animals are capable of performances that resemble human ones (or, as Leibniz put it, that “imitate reason”), and which indeed we use 99% of the time. A dog recognises its owner, and does so whether or not they are wearing a hat, whether they are sitting or standing. This does not necessarily mean that the dog has a concept of its owner (just as it seems exaggerated and pointless to assume that the dog, when it sees the owner, formulates the judgement “here is my owner”), but it certainly does have a schema that it uses for reference. There is therefore a schema of “dog” in the subjective sense of the genitive. Then, there is a schema of “dog” in the objective sense of the genitive, which is what Kant talks about: how does the owner recognise their dog? It’s obvious, they know it, it’s their dog, they’ve already seen it. But, then, how can you recognise a dog (any dog, therefore a dog in general) if you have never seen one? The word “dog” is useless, one needs something like a figure. Heidegger – who said that animals do not die but perish, do not think, and do not even have hands, but paws – said that thinking is talking. Monkeys, who don’t talk, therefore wouldn’t have hands. It’s hard to tell where he could have seen a monkey in the Black Forest, but it probably was a figure in a book.

2Now, things get more complicated with less obvious animals, say the dugong or the platypus. How do you classify a platypus? Can there be an a priori platypus? Returning to the classic dog, there is a central point in the Critique of Pure Reason: «the concept ‘dog’ is a rule in accordance with which my imagination can specify [verzeichnen] the shape of a four-footed animal in general, without being restricted to any single particular shape that experience offers me or any possible image that I can exhibit in concreto» (B 180/A 141). The problem is that we can’t see why, talking about the pure concepts of the intellect, Kant should talk about an empirical concept like that of a dog. And the somewhat widespread suspicion, since the early days of Kant’s reception, has been that the empirical should have, in the case of schematism, a transcendental function, and almost serve as a model for the pure concepts of the intellect. Umberto Eco, Diego Marconi and I were amazed to find in our three books published in 1997 – respectively, Kant and the Platypus (Kant e l’ornitorinco); Lexical Competence; Estetica razionale – this possibly inevitable topos, together with discussions of other animals (platypus, dugongs, dolphins, and rabbit-ducks).

Diego Marconi - The Black Box

  • 1 Or of the underlying concepts. Here the distinction is unimportant.

3Concerning the dog schema, this is how I see it: normal competence on the word ‘dog’ includes the ability to answer questions about dogs, the ability to realize that a norm that concerns animals applies to dogs, the ability to grasp the inconsistency of “All dogs are platypuses”, and more. All such abilities depend on overall mastery of a network of words1 that I call ‘inferential competence’, as it grounds the ability to perform so called “semantic” inferences. Moreover, normal competence on ‘dog’ includes further abilities, different from those I just mentioned and irreducible to them: those which make it possible for us to apply the word ‘dog’, i.e. to buy a dog, not a platypus, if we promised our daughter to buy her a dog for Christmas, to draw something that resembles a dog (while being quite different from a platypus) if we are asked to, and so forth (referential competence). For such abilities, most particularly for the naming ability – the ability to call a dog ‘dog’ – ability to recognize dogs is crucial: i.e. the ability to analyse the perceptual scene by way of procedures that (allowing for errors, distortions, etc.) succeed if a dog is present in the scene, and fail otherwise. I.e., the word ‘dog’ is associated with a recognition algorithm.

  • 2 Marconi 1997: 79-82, 145-149.

4For reasons I spelled out at some length in my book Lexical Competence,2 it is essential that procedures are involved here rather than images as in the empiricist tradition. This, I believe, is the central intuition of the Kantian doctrine of schematism. However, Kant had identified only one reason of the inadequacy of the empiricist theory of recognition: that a concept’s application conditions cannot be implemented by an image, because every image is exceeded in generality by the corresponding concept. In other words, the application of the word ‘dog’ cannot hinge on possessing some image of a dog (not even an image of the typical dog, in case that makes sense), for the image’s salient properties involve an undue restriction with respect to the word’s application conditions: e.g., the dog image may be one of a long-haired animal, though ‘dog’ also applies to short-haired animals. Thus, images fail to capture the generality of concepts.

5There is, however, another, equally important reason of inadequacy, which is nowhere to be found in Kant’s considerations. The problem – which, in our times, was highlighted by Wittgenstein and later again by Hilary Putnam – is that an image does not include its own application. That the word ‘book’ is associated with a certain image (of a book, or of the typical book) doesn’t tell me what it is for that image to correspond to my visual experience, or to part of it; in fact, it doesn’t tell me what it is for the image to correspond to anything whatever. Images, in themselves, are just pictures, and pictures do not correspond to anything: we have them correspond to something. Hence, for the purpose of recognition, replacing a single image by a general method to generate images, as Kant suggests, is not enough: recognition of objects (or actions, etc.) does not just require that some image is generated supposedly corresponding to some object, or to our perception of it; it also requires that the image is itself recognized as corresponding to the object. More to the point, correspondence of image and object cannot be taken for granted: one ought to specify what it is for an image to correspond to an object. When cognitive structures are involved, there have to be procedures that explicate the idea of correspondence (A corresponds to B if and only if such and such procedures associated with A succeed when applied to B).

  • 3 See e.g. Paternoster, Meini 1996: 1-2, 37-48.

6So, to account for recognition we must suppose that what is associated with a word such as ‘dog’ is not just a general method for the production of images (which images we wouldn’t know how to apply to the output of perception), but a method for the analysis of perception. The “dog schema” we need does not generate images of dogs; it integrates (in several different ways) salient perceptual features. Each such integration is the recognition of a dog. I have no idea of how that takes place in our mind (not to mention our brain); however, artificial vision shows – within limits – how that can take place in a computational system.3

  • 4 Ferraris 1997, § 3.6: 344.

7Having thus offered my own view, let me add one remark for Maurizio Ferraris and one for Umberto Eco. Right at the beginning of his paragraph on the dog schema, and almost in passing, Ferraris says: «Schematism is a case of seeing-as, and seeing-as is not a perceptological exception; it is just normal seeing [la semplice norma del vedere]».4 Anybody familiar with Wittgenstein will imagine why I have been struck by this claim: we have been taught, both by Wittgenstein and by the many discussions concerning Chapter XI of his Philosophical Investigations, Part II, that we should draw a distinction between seeing-as and simple seeing: not every seeing is seeing-as. In certain circumstances, we say that we see a diamond as a square in perspective, or that we see two concentric circles as a Mexican hat see from above; we do not say that we see a knife as a knife, or that we see our dog as our dog. Does the distinction only concern our use of language (this is what we happen to say, but, according to Ferraris, we ought to say otherwise)? I don’t think so; or at any rate, as I see it, the distinction can be drawn in a simple way as a cognitive, not just a linguistic distinction. In the seeing-as case (but not in the simple seeing case) there are two recognition procedures, both of which succeed, even though one is somewhat privileged – for example, because it is the first to be run: I first see the diamond (I recognize a diamond), but I can also see the figure as a square in perspective. Hence the interest of such cases as Wittgenstein’s duck-rabbit, and perhaps Ferraris’s son’s puppet, where there appears to be no privileged recognition procedure.

8At the cognitive level, it is not the case that every seeing is a seeing-as. What does Ferraris mean, when he describes schematism in general as seeing-as? That every perceptual (visual) act has a determinate content (seeing a dog), or that every perceptual act is an act of recognition (identifying a dog as such in seeing it)? Both alternatives are hard to make precise, and they appear to involve intricate debates on the non-conceptual content of perception; more generally, on the relations between language, concepts, and perception. Too many things, and much too hard, to be dealt with on the present occasion.

  • 5 Eco 2000: 123.

9For Eco, I have a general methodological problem. Eco declares that he does not intend to «poke [his] nose into the black box of our mind or brain processes».5 It seems to me that if one shares Eco’s theoretical goals, one ought to stick it right there: indeed, some less convincing aspects of his book may depend on insufficient interaction with psychology and the neurosciences. I would like it to be clear that I am not arguing that any theoretical work on language must involve such interaction. I am arguing that such interaction is needed if one intends to put forth hypotheses about cognitive processes, whether they involve language or vision or arithmetical reasoning. Otherwise, one ends up talking not about cognitive processes but about the structure of certain discursive domains (e.g., not about arithmetical reasoning but about arithmetics), while interspersing the theoretical scene with purely speculative entities and processes, that are attributed to the mind without trying to substantiate them with what we know about the mind; in other words, one ends up doing parascience, to borrow Putnam’s term.

  • 6 Cf. Eco 2000: 131.
  • 7 Cf. Eco 2000: 155.
  • 8 In my book Lexical Competence, I try to show that the neuropsychological data suggest that applicat (...)
  • 9 Eco 2000: 168.
  • 10 Ibidem: 163.

10This, in my opinion, is what happens with Eco’s notion of a cognitive type (= CT). CTs are initially introduced as cognitive structures that are in charge of recognition of objects such as horses6 (in my vocabulary, they mediate our referential competence with words such as ‘horse’). Later,7 however, they are also attributed the function of producing images or pictures, and there is talk of “constructing and identifying” a picture.8 Moreover, CTs do not just mediate referential competence by way of shape recognition; they also include information that make it possible to identify something based on its actions («The bishop in an ordination ceremony and a referee (even if in plain clothes) in a football match are recognizable by what they do»);9 such identifications require «a reference to a framework of cultural rules».10 There’s more: there are CTs for each perceptual modality (e.g. acoustic CTs, allowing us to recognize the Fifth Symphony), as well as individual CTs (e.g. a CT for my wife, not just one for women in general).

11Now, there is no doubt that in every case Eco mentions (referees, the Fifth Symphony, my wife, horses, etc.) it is appropriate to speak of “recognition”. However, this does not entail that we can aptly postulate a single kind of cognitive structure – CT – that would be active in each and every such case of recognition. Postulating such a structure in the lack of any psychological confirmation would amount to projecting “into our head” the common usage of the word ‘recognize’. It is a bit as if one argued that there must be an organ in our body, the “vital resource processor” (VRP), which is in charge of extracting from the environment oxygen, glucose, and other useful substances to eventually enrich the bloodstream. Though it is true that our body extracts such substances from the environment and that they end up enriching the bloodstream, no single organ performs that function – indeed, there is no single function either. The VRP is the speculative hypostasis of a number of functions, which are best identified and described separately, for a thousand of good reasons: they evolved separately, they are implemented by materially distinct and separable organs, they are partly dissociable, etc. It is important to notice that, to realize that there is no such thing as the ERV, anatomy is not essential (though it does provide useful hints): it is enough to realize that, at least for some time, we can breath well while having bad digestion, or the other way around.

  • 11 See Marconi 1997: 68.

12Similarly, Eco’s cognitive types are the hypostasis of a number of distinct cognitive functions, and to realize that that is the case we don’t need PET. It is enough to realize that one can be entirely able to call a horse ‘horse’ while being totally unable to draw a picture of a horse or even to imagine one,11 or that one may lose the ability to associate a shape with a function or class of functions while preserving the ability to recognize that shape; and so on.

Maurizio Ferraris - Straightening Dogs’ Legs12

  • 12 Italian saying meaning “attempting the impossible” (translator’s note).

13Is it really the case that we either say exactly how things are, or we should remain silent? If it were so (which cannot be ruled out), I would be forced to quit the game right away. And yet, even without being sure of having found a keystone or a definitive answer, not all the answers are the same – for example Kant, as Marconi recalls, illustrated how reference works in a more satisfactory way than many contemporary theorists, even if he ultimately made everything depend on a mystery placed in the depths of the human soul. Should he rather have kept quiet?

14What seems interesting to me in the schema ‘dog’ is the fact that when dealing with the schematism of the pure concepts of the intellect Kant comes to speak of the schema of an eminently empirical concept, that of dog, that is, something that in § 59 of the Critique of Judgment he would call an example. As we know, the schema in Kant falls within a complex field which includes three terms: the schema of empirical concepts (the example); the schema of pure concepts (the “real” schema); the schema of ideas of reason (the symbol). Strictly speaking, one should also consider a fourth type of schema, the number, which however does not have, at least from Kant’s perspective, a residual matter, so it can proceed by pure construction and is not a hypotyposis. This brings up a topic that is too broad to address here; but, sticking to the typology of the Critique of Judgment, I think it can be said that none of the three canonical types of schemas can prescind from the empirical.

15This is very clear in the example and the symbol. In the first case, a given image is valid for a whole class (or even for several classes: example of dog, example of Doberman, etc.); in the second case, the sensible signals towards an idea of reason of which we have no clear and distinct idea (it is only a vague teleological principle) and which by definition exceeds any given representation. We really do not know what loyalty is, that is, what it means exactly to be loyal, while we know what substance is; so I can symbolise loyalty, for example, by taking the same Doberman as before, and taking it as an example of loyalty from which humans would have much to learn. I would like to point out that in the case of the example, as well as in the case of the symbol, the same motifs that can be used to demonstrate that the concept exceeds the image can also be used to demonstrate that the image exceeds the concept; in fact, the Doberman in question is not only valid as an example of caninity, but also as a symbol of fidelity. All that remains are the actual schemas, i.e. those concerning the application of the pure concepts of the intellect. Now, it seems to me that pure concepts should simply be assumed, as nothing is in the intellect that was not previously in the senses, except the intellect itself (i.e. the possibility of retaining, that which Kant designates as the synthetic unit of apperception); which also means that every pure concept schema involves its application simply because it is already applied, and is obtained by abstraction (giving life first to the judgement, then to the category; Hegel already ironized about the twelve categories that Kant finds “ready-made”, and although Kant blames Aristotle for finding the categories “rhapsodically”, it is difficult not to think that Kant’s categories are largely the same as Aristotle’s, not found rhapsodically only because they were taken from Wolff). I see wax solid and fragrant, then I recognise that it is still wax when it is liquid and odourless; but against Descartes’ example one could oppose Locke’s example of the king of Siam, who did not believe that ice was solid water because he had never seen it in his country. One might say: these are simple empirical examples; but, in fact, I challenge anyone to find a concept without an example, that is, without an image (we even have an image, albeit very inadequate, of a square circle or of nothingness).

16If this is the case, “my” hypothesis is that the notion of example, i.e. of a single self-transcending case, is the common root of the schema of dog in both the subjective and the objective sense, and that what Kant says about schematism can be traced back to what in Estetica razionale I called «exemplarity of the example». The use of the word ‘example’ is significant from this point of view, and if I recall it briefly it is not in the conviction that its etymology holds some kind of secret, but because I believe that the use of a word is phenomenologically significant. On the one hand, the example is something very common, the example of an object; on the other hand, it is extraordinary, as in the expression “setting an example”. This is the same duplicity at work in the words ‘type’ (and ‘typical’), and ‘model’. So, for example, the German word Muster means model understood as a facsimile, but it is at the origin of the word “monster”, precisely because this exhibition, domestic as it may be – for example, the image of an ATM used to illustrate the procedure to follow – is in any case monstrous, i.e. it shows not only itself, but also something that exceeds it. The ATM Muster would be, in the Kantian lexicon, a hypotyposis, and precisely that kind of exhibition of a subjectum sub abspectum which concerns empirical concepts. But to what extent does something empirical remain such if it exceeds itself? In the Metaphysics of Morals, Kant underlines the difference between Beispiel, as a case of a law, and Exempel, as a case that, in a reflective way, can give birth to a new law. But, in fact, it is not easy to distinguish the ordinary from the extraordinary, as well as the determining judgement from the reflective judgement. Something is both a case of a given law, subject to a determining judgement (“here is a pen”), and a recognition schema (“look for a pen like this”). In an even more interesting form, in order for a single case that has not yet been classified to become the principle of a law, it is necessary for it to have access to some kind of exemplariness; this is what Kant says of works of art (the single case that stands at the beginning of a canon, e.g. a literary masterpiece as opposed to genre fiction); but it could be very well applied to the platypus.

17Now, if we try to recognise what holds all these performances together (being an ordinary case, being a recognition pattern, being the starting point of a law) we will see that it is an essentially teleological dimension. The example lends itself to imitation, or to that specific form of imitation which consists of application (I will return to this point in a moment). To play with language a little bit, in the example we find the ambiguity of the medium (which is the middle part, a substance, and a means to an end) or of the element (which indicates an atom, a part, and the ether, the context within which this part is given). But, to return to the classic problems of gnosiology, it seems to me that the example has the characteristics of singularity and universality that we find in the “diagram” with which Berkeley proposes to solve the problem of general ideas left unsolved by Locke, i.e. in the single case associated with universality. And, as I stress several times in Estetica razionale, there is a short gap between the diagram and Kant’s monogram, i.e. the schema. The schema is in fact, at the same time, the representation of the structure of an individual without its more or less accidental peculiarities (for example, the schema of book) and a structure predisposed for an action (for example, the schema of a coach for his football team). In this sense, and again I don’t suppose this affinity is purely accidental, we are dealing with something that is both a precedent and a model for an action to come, as happens in the Italian notion of “traccia” (track or outline), which is both what remains of something (the traces of a civilization etc.) and something prepared for later execution (as when we say “the outline of a speech”). Now, by far the most interesting aspect in the example is that the excellence of its exemplariness consists in its banality, in being extraordinary because it is absolutely normal, so that an abnormal example would not be a good example at all. This is the reason why Kant maintains that Christ should not be taken as an example of moral life (because then everyone would be inadequate, and absolved); and it is the reason for the monstrosity of the platypus, which seems to be made from pieces of other animals, so whereas it should embody the “normal idea” of an animal, it embarrasses zoologists instead. Indeed, it forces them to go from a determining judgement to a reflective judgement (or perhaps to show that any determining judgement is but a reflective judgement whose origins have been forgotten).

18The case of the schema of dog in the Critique of Pure Reason can only be explained by the multiplicity I have tried to sketch, otherwise one would not understand in what sense (and it is certainly a slip of the tongue, at least with respect to the explicit structure of Kant’s doctrine) Kant can assert that I have both the schema of substance as the permanence of something in time (i.e. of an a priori concept that for him must be presupposed as prior to experience) and the schema of dog (i.e. of a concept that, unless we presuppose a system of innate ideas much more detailed than in our most daring dreams, can hardly be prior to experience). And, here too, we are dealing with a certain duplicity. On the one hand, I cannot have the concept of purely a priori substance, i.e. before the experience of any substance. What Kant says about the laws of the intellect should in fact refer to the laws of the senses, that is to say to the fact that, without having read Aristotle, I still recognise a substance in the firm and apparently solid thing that is in my hands or under my nose. So much so that I may have all the contrary philosophical beliefs in this world (or, which is more significant, I may have no philosophical beliefs at all), and the senses will invariably tell me that there is a substance, to the point that I will have to devote myself, for example, to transcendental meditation to convince myself that the substance is mere appearance, that is, a deception not of the intellect, but of the senses. In other words, if the principle is a priori and transcendental, it is only such to the extent that it concerns the senses and not the intellect (provided that this distinction really holds), then – and Kant acknowledges this – it arises together with experience, even though it must be assumed as prior to it.

  • 13 I would like to provide a few supporting elements. Apparently, there is nothing more distant from t (...)

19So, for example, in drawing a line I have both the line and the unity of consciousness in the concept of line. I do not believe that this circumstance is a dismissal of the transcendental, and it seems to me that indeed it is the only condition for an authentic transcendental. A possible split between a priori and a posteriori would not guarantee the purity of the a priori but would make it something determined by the a posteriori, therefore something purely empirical and constituted. A transcendental that was purely a priori and analytic would not be a purer transcendental – it would not be a transcendental at all, it would be merely empirical. By not determining experience (or at least not extending it), it would actually appear as determined by experience, i.e. it would be placed in an already established area (it is a fact that bachelors are not married). In other words, the “quasi-transcendental” we are talking about under the title of the exemplarity of the example would ensure the only possible constitution of the transcendental.13 This consideration explains two points, I think. First, how there can be a schema of empirical concepts, and in what sense the example is the principle that can also constitute pure concepts of the intellect, and not the other way around. Secondly, one also understands in what way the schema of dog can be given in the subjective sense of the genitive, namely in what way dogs, who perhaps do not have concepts, certainly have schemas, that is they have a perception that is not reduced to seeing, but also involves seeing-as, i.e. it allows them to recognise a hare in various ways, whether stationary, moving, or skinned in the kitchen.

20Having made these prolonged premises, I will try to respond to Marconi. As for “seeing as”, I certainly do not want to straighten the legs of dogs, nor do I delude myself that I can reform the vocabulary or the Italian language by suggesting the cumbersome replacement of “seeing” with “seeing as”. However, I would like to observe that the fact that I do not say “I see that dog as a dog” comes from the same reason why I do not say “there is a dog on the bottom of my retina” but rather “there is a dog in the courtyard”. If I wanted to (if I was very accurate or, better, pedantic), I could very well say “I see an x as a dog” (which moreover, for lack of zoological culture, I cannot specify as a Dachshund or Doberman). This would be the situation illustrated by Austin in Sense and Sensibilia: only someone pedantic (or, more exactly, a skeptic) would say “it seems to me that that red flower can be considered, under certain conditions, a geranium”, or “I see the image of Sirius in the twelfth mirror of the telescope”. Rather, I simply say “that’s a geranium” or “I see Sirius”. Now, just as it is not difficult, if one is conscientious, to go from “there is a dog” to “I see a dog” (which involves some skepticism: “it seems to me that there is a dog; that it really is a dog, I don’t know”), so it is not difficult to pass – if one is pedantic, or a transcendental philosopher – from “I see a dog” to “I see that as a dog”; it depends on the attitude one takes, which in the first case would be that of sensory certainty, in the second that of methodical doubt, and in the third, precisely, that of transcendental philosophy. I am not so much worried about the objection that if we assimilate “see” and “see as” we will never be sure to see, and we will always suspect that we are seeing as. Just as we empirically know how to distinguish, generally and without difficulty, between seeing and seeing as, I do not see anything wrong in transcendentally identifying vision and vision-as (in the same way as, ideally, perception and matterless form are not distinguished, since in both cases it is about eidos aneu tes hyles.).

21The question I would like to raise with Marconi, especially with reference to the considerations made on pp. 145-149 of Lexical Competence, is placed somehow the other side of the reasoning conducted so far. The point is this: are we sure that the (empirical) image does not include its application (i.e. it is transcendental)? To what extent is Kant evasive on this aspect? What exactly does it mean to explain the application of a schema? There are two problems here, and they concern the application and the image.

22As for the first aspect: are we sure that, speaking of application, we are not confusing two instances, the one that (for example) explains the learning of the mother tongue and the one that explains the learning of a second language? First we are in a situation where concepts and perceptions coincide (this is the stage to which Aristotle alludes at the beginning of Peri hermeneias when he says that words are “equal to things”), then we explain how, starting from this schema or from this common root, concepts and perceptions can be defined. This is a problem which has to do with the mystery laid down in the depths of the human soul, and which cannot be explained, in my opinion, by looking into the black box, because you would find exactly what you put in it, but which can be described and refined phenomenologically. Then there is another very different problem, that of how, starting from this primary competence, one can, e.g., explain what a computer looks like to someone who has never seen one (one can say that it looks like a typewriter plus a television) or draw a map to get to a party in the countryside, etc.. This is generally the difference between learning the mother tongue and another language.

23The ambiguity of schematism is that on the one hand, in the style of the studies of that time on the origin of languages, it explains how de facto we refer to the world (and this concerns the function of the schema as subsumption), and on the other hand it explains how de iure we refer to the world without falling into Hume’s doctrines (and this concerns the function of the schema as constitution of experience). Now, if we take this second version, Kant should explain schematism by saying that, for example, I first think about the concept of substance, then I take the schema of the permanence of a thing in time, and finally I apply it to my table. But what happens, according to the first version, is that when I touch my table, I activate both the perception of something that resists and the concept of substance. In a way, to raise the problem of application is rather untimely, because the application has already taken place, and to wonder how the application happens means wondering, as we speak, how we should make sounds in such a way as to speak. It should be noted, however, that even with an apparently more empirical version (schemas are born with experience) it still has not been said that schemas are born from experience, or identify with it, and here it seems to me that Kant’s opposition to Hume maintains the same validity as Leibniz’s objection to Locke (nothing is in the intellect that was not in the senses before, except the intellect); empirical and transcendental are two parallels, i.e. they are not identical.

24As far as images are concerned, I am not at all sure that they are necessarily more determined than concepts (while it seems to me that they are less determined than perceptions, or, more exactly, than perceived individuals). The image can be an object of contemplation (theorem) and a tool for memory (mnemoneuma), and this means that it is at least two things. When you write that “images, in themselves, are only figures, and figures do not correspond to anything: it is we who put them in correspondence with something”, you are saying something very true, but perhaps a little broad and platonic; isn’t that what Socrates says to Phaedrus, namely that images appear alive, but when you question them they remain silent?

25Socrates thinks that the true logos is the one written in the soul of the learner, and that it knows whom to answer and for whom to remain silent, that is, he supposes that on the one hand there is an absolute pole of intentionality, and on the other hand the sphere of inert things and instruments. But are we sure, first of all, that all things can adapt to become instruments, that each instrument is like any other, that it cannot determine its end at least as much as it is determined by it (for this reason the means justifies the end at least as much as the reverse), and that on the one hand there is an absolutely intentional pole and on the other an inert world? So why do we say e.g. that opportunity makes a thief, or that si vis pacem para bellum is a hypocritical motto because once the arsenals are full then war will be waged? If you have a car you end up wanting to go to the seaside, remaining stuck in traffic for hours, and keeping away from temptations (which are actually intentions) means keeping away from the means to realize them. One could also argue that cars are useless, and it is we who drive them; but that does not mean that you can drive a pumpkin in the same way you drive a car.

26Returning to our problem (though it is clear that the schema has an eminently technological character), I would be personally less harsh towards images, than with, for example, bananas as such; the former, in fact, have the characteristic of being not only things, but also images of things that can be used technologically (even if with a verbogene Kunst), in the same way as a branch can also be a lever. Take the typology that Heidegger proposes in Kant and the Problem of Metaphysics: the photo of a death mask shows many things: the face of the deceased, what a death mask is in general, and what a photograph is in general. This consideration comes directly from the analysis of the symbol in Hegel: the image of a lion can represent this lion, a lion in general, as well as strength and royalty. And here we can return to the question of application, where I would like to say that images contain the rule of their application – to be clear, the image does not teach us how to apply it, but tells us that we must apply it, because it refers to something that does not exist as such, and I would exclude that there is only one way as for “how”. Personally, however, rather than images, I would speak of traces, i.e. impressions that remind us of previous experiences, stimulate analogies, call to mind procedures, which in turn can be extraordinarily different depending on the objects, subjects, and even circumstances (if, for example, there was only one way to temporize our experience, there wouldn’t be the phenomenon by which time flies or stops).

27I now come to the question for Eco. It seems to me that in Platypus the convergence between schematism and exemplarism is fully ascertained, indeed it is somehow the theme of the book. What you say about schematism, in fact, is essentially addressed to the constitution of applied schemas, i.e. examples: you maintain that if you had to rewrite A Theory of Semiotics, you would do so starting from the modes of sign production, and that the schema fluctuates between the immediate object and the cognitive type (which is precisely the exemplariness of the example). And the apologies you make are essentially referred to schemas of empirical concepts (not only the platypus, but Marco Polo’s rhinoceros, the horses seen by the Aztecs, etc.).

  • 14 Eco 2000: 52.

28In 3.6 you do not take a stance on the question of the chicken and egg, i.e. on the question of which came first, schemas or things, because it is a mystery that, even if solved, would not clarify much. In the same way, the ontological role of language is strongly diminished, and in several decisive points. But then why do you write, for example, that «being always manifests itself in language only»?14 (I could mention other occurrences). I would not say so: I would say that being always and only manifests itself as a form, that is, as a presence (sensible or intelligible), and it does not seem to me that language intervenes so decisively in the constitution of the morphè (unless it is used to indicate any system of inscription and retention, and I do not see why one should do that).

  • 15 Cf. Eco 2000: 24.

29In short, my question, in its most rudimentary form, would sound like this: on more than one occasion you seem to argue that being has meaning only when it is said in words, but this is something difficult to agree with, because meaning is a wonderful word, and one of its two senses (it doesn’t matter whether it is the first or the second), the one I get by hitting a wall, doesn’t require words. As always happens when language comes into play, this is a bit like, while having failed to unravel the mystery of schematism, holding on to the idea that there is, in any case, a sphere – language – in which we are masters of ourselves instead of slaves to a referral and reference mechanism of which we understand little. But it is not necessary to know how hands work to use them properly, freely and responsibly. And, if that’s the case, I don’t see why we should give so much importance to language. This seems to me a remnant of a “logocentric” perspective, whereby signs are an expression of consciousness (or, more precisely, the sign is but a passage or a catwalk between the presence of the object and the presence of consciousness to itself). But couldn’t we say the opposite? On closer inspection, there is not a single reason to think that there is an intention prior to retention and that it comes to animate the sign as the spirit animates the body. For example, who has ever said that it is the spirit that animates the sign by applying it? Well, the list includes (but not without problems and contradictions), Plato, Hegel, Husserl, and, at least in part, Marconi when he says that it is we who use images, and yourself when you write that being is only an effect of language15 (even if you then observe that language is insufficient to define it). But is it necessary to say this? Is it not possible, rather, that it is the sign that gives rise to the spirit? In phenomenological terms (at least in Derrida’s revised phenomenology), addressing is a sign that precedes consciousness and also language.

30One can say that, in the end, not much changes in these very primal things. I, however, am not convinced. Something does change, if not from a theoretical point of view, at least from a moral one. Giving priority to an ideal intentionality and a conscious language leads us to flatter ourselves (as happens in Heidegger: “only man speaks, only man thinks, only man has a hand, only man dies, but the animal, even if it could laugh, would have little to laugh about, because it perishes”); and, therefore, it makes us much more immaterialistic than we would like. That is to say, it leads us rather perversely to think that words, so comfortable and at hand, “make things be” (and maybe even make them disappear). In short, the reference to language over-excites us, because it makes us depend too much on the spirit. But it is one thing to think, another to speak, another yet to perceive. One of the best parts of Platypus is when you show (with the example of Australians) how little the categories have to do with grammar, which ends up having devastating effects on cultured beliefs about the fact that the subject/predicate structure determines (or is determined by, everything goes) the substance/accident structure, so whoever has language has the key to all things. On the one hand, this is an explanation that proves too much (it is like saying that the book of nature is written in mathematical language, and there’s that, without making a single experiment; in which case it is like saying that whoever knows the letters of the alphabet knows all that is contained in books). On the other hand, it’s an explanation that doesn’t explain anything in detail, because, if it was really operative in a strong sense, maybe a Hopi should not get hurt when he slips, because in his language there is no subject-predicate structure, so he shouldn’t crash against those annoying synbebekota that are on the ground, and that hurt like hell, because they are attached to a sturdy hypokeimenon.

31Speaking of examples and schemas, traces and signs, can one explain anything more? In part, yes (it is the sense of the perfectibility of human reason to which I was referring at the beginning), because, for example, one can explain how, without any reasoning, we are able to walk and dodge obstacles, and why a beaten dog flees (exactly as his master would do in the same circumstances) at the sight of the stick without having (I believe) concepts and without having (and of this I am more sure) words. If then you think that this is still too little, well, what’s the advantage of believing that language is, if not the house, at least the abode of being?

Diego Marconi - Compasses to Chimps

32I am not sure I understand Ferraris’s first remark. It appears he charges me with lamenting that Kant does not explain how schemas are applied (“What does it mean exactly to explain how a schema is applied?”). If so, then he misunderstood: in the pages Ferraris quotes, I claim that Kant is insensitive to one of two reasons why images – not schemas – cannot mediate between language and perception, the reason being that they do not come with rules for their own application. Kant justifies the procedural nature of mediation – i.e. the procedural nature of schemas – on the single (Berkeleyan) ground that images are particular whereas concepts are general, while I claim that the crucial motivation ought to be a different one, which Kant does not mention.

33Concerning Ferraris’s second remark, I am inclined to think that our difference may depend on his taking some (conjectural) features of our species for granted, whereas my argument is meant to be more general. Let me explain. It may be claimed that there are natural resemblances: e.g., a picture of a pear looks like a pear. What could that mean? For example, it could mean that any being that is more or less like us recognizes the picture as the picture of a pear (not of a sofa or of a vervet). In other words, it is quite possible that we are so shaped as to spontaneously grasp certain correspondences, with no need for special training. Maybe such an ability is an evolutionary advantage we have, like being endowed with language. By contrast, I don’t know what it would be for an image to be inherently a picture of this or that. To see that there can’t be any such thing, it is enough to consider that if we provide a computer with a set of images, and further endow it with a system of artificial vision, it won’t have any use for such images until it is taught how it is supposed to use one of them to analyse the scenes it accesses thanks to its vision module (i.e., until it is provided with application procedures). The image in itself does not tell one how it is to be applied, or that it has to be. Of course, it tells us both things, particularly in the case of images that are culturally entrenched, that conform to certain picturing conventions, and so forth, and perhaps even in the general case – supposedly, because of the kind of beings we are.

34Concerning this point, let me add that comparison with tools seems to me to comfort my view. Once something has been identified as a possible instrument for some purpose, that it is a better or a worse instrument (for that purpose) depends on its make, not on us. However, this doesn’t mean that it is “inherently” an instrument for this or that purpose: try to provide a three-year-old chimp with a compass. Similarly, though there are both good and bad pictures of a pear, they need first of all to be dealt with as pictures (not as coasters for glasses of Coke), i.e., the category of resemblance has to be applied to them and it has to work in a certain way, i.e. by determining a certain analysis of perception, etc.

35Let me further add – as I am favored by the discussion’s format – a reply concerning seeing and seeing as. The issue is not whether simple seeing could be redescribed so as to make its conceptual nature explicit; the issue is that we ought not to lose the distinction between simple seeing and genuine seeing as, e.g., between seeing a fork and seeing two concentric circles as a Mexican hat seen from above, or seeing Jastrow’s image as a duck rather than as a rabbit (the cases are different, but that is nor relevant here). If simple seeing is collapsed on seeing as, we risk losing the distinction: what, then, is genuine seeing as? Yet it looks like a useful distinction. It can be challenged, of course; however, it doesn’t seem to me that this is what Ferraris is up to.

Umberto Eco – Dogs and Sea Spaghetti

36I am not saying anything new by observing that every time two or three philosophers (and even non-philosophers) get together to ask themselves how it happens that we know and if there is a world to know, they get involved in a series of issues from which they don’t know how to get out. So, just as the old Thomists used to say that there is no progress in metaphysics, I think we can say that there is no progress in ontology or in gnoseology. It is enough to understand the expression “there is no progress in...” not as the old Thomists did – i.e. that Thomas Aquinas had already answered all questions – but in the exact opposite way, namely that no one has yet given a satisfactory answer and therefore the interesting questions are still the same.

37This is to say that if I wanted to get involved in all the issues that my two interlocutors have been stirring up (since I write after having read them – for accidental reasons, we ended up applying the academic rule by which the oldest speaks last) I would only increase the size of the quagmire that they have created (I would say more so Ferraris, who played offence, than Marconi, who remained more cautious). I will therefore limit myself to responding to objections addressed directly to me, rather than to the World. And with a further act of prudence, I wish to recall the origin of some of my monothreematic arguments (yes, I mean it spelt like that) that to some extent hover over this conversation from a distance.

38This conversation did not arise, as it may seem at first, from the fact that the three of us have, almost at the same time, brought up the issue of Kantian schematism, because we would not be the only ones – nor this fact would be accidental, if it is true that notions very similar to that of schema circulate abundantly within cognitive studies. I would say that the problem we have in common, perhaps due to some remote common origin, and beyond the philosophical parties in which we have been involved (as Ferraris reminded us on page 20 of the last issue) is that even if it were true that there are no facts but only interpretations, eventually we will have to say what it is that we are interpreting, and as soon as this something is mentioned, any ontology thrown out the door comes back through the window.

39This would be true even if everything that we believe to be a fact was already known to us due to the mediation of a text. A text is also a fact. I am taking up an old argument (from The Limits of Interpretation) against the idea that every interpretation is a misunderstanding: given two texts Alpha and Beta and an interpretation Gamma, is it possible to decide if Gamma is an interpretation of Alpha or Beta? If it is not possible, there is no interpretation, but only production of texts with no relationship between them, pure solipsistic stammering. If instead it is possible, then there is a parameter to discriminate reliable interpretations from unreliable ones (because to say that Gamma is not an interpretation of Beta it is necessary to say that Beta is not the Thing we are talking about). Now it is true that, reasoning in abstract terms, it is also possible to argue that Gamma could be the interpretation of both Alpha and Beta, but in the communicative trade things are much simpler: is Monti’s Iliad (a very free translation of other translations, as we know) the translation of Homer’s Iliad or Virgil’s Aeneid? Let us not be foolish, it is very possible to give an answer, and anyone who says otherwise would not be taken seriously even in the French department of an American university.

40This long introductory parenthesis (I hope you appreciate the oxymoron) was meant to explain, among other things, why such a debate should take place in an aesthetics journal. Both those who reflect on the philosophy of language and semiotics, and those who reflect on aesthetics evidently share the idea that etymology should be taken very seriously, and that aesthetics, although it may concern the problem of beauty or art, in the first instance also concerns the moment when something is perceived (which is no small matter in a time where so much has been said about purely conceptual art). This is an interesting return to the primality of perception, by people who have long been engaged on different fronts, a sign perhaps of a Zeitgeist – and most likely the right reaction to a sort of oblivion of the multiple of intuition, after two centuries of general idealistic-transcendental drunkenness.

41After this premise, I wish to immediately respond to a first question posed by Ferraris, the most decisive one, because it seems to me that it hides a sort of overreaction to the idealistic-transcendental binge. Why give so much importance to language? It’s one thing to think and speak and another to perceive, a dog probably doesn’t think, certainly doesn’t speak, but it certainly perceives, recognizes its owner and feels pain if it hits the wall...Well, a first answer requires clarifying the genesis of my Kant and the Platypus, which was not born as an organic treatise but as a series of problematically vague reflections. The essay on being, which opens it, is from 1994, therefore three years earlier than the others. Naturally, the fact that on one day in 1994 I decided to reflect on being also determined my decision to write the other essays. And if, on an ontological level, I decided to say that there are lines of resistance in being (the “hard core”) this was bound to influence my semiotic conclusions about it. But the semiotic conclusions I drew from these reflections are, so to speak, methodologically independent from that wonderful revelation. I mean that if you read in my first essay that being is an effect of language, you can see that I am saying this when talking about the classical history of reflection on being, from Aristotle onwards, and I am talking about two very precise questions.

42One is that being becomes a problem, with Aristotle, when one realizes that (in common discourse) it is said in many ways (it is enough to keep quiet and the problem would not arise); the other is that being as a philosophical problem, the idea of being as being, arises only as the effect of a question, why is there being rather than nothing – or other such questions. I was the first to say, and right at the beginning of that first essay, that the experience of the fact that something is there precedes every philosophy and every language, is available to animals, perhaps to foetuses, and certainly to newborns, and I used as an authority my dearest teacher Thomas Aquinas: being is “quod primum intellectus concipit quasi notissimum…”. But if this experience remained such (unexpressed) we could only talk about it (i.e. formulate our refusal to talk about it) in a “Journal of Ecstatics”. And in a “Journal of Ecstatics” we could refuse to talk about that wonderful experience (according to Ferraris) that one has when hitting one’s head against a wall. What should we do to go beyond this mute experience (interjections aside)?

43I do not know how dogs think. Marconi reproaches me for not wanting to stick my nose into the black box of humans, let alone in that of animals. But I certainly do not refuse to make thought experiments on the subject – indeed I often make them in free zones, and I would like to remind readers that in my latest novel, The Island of the Day Before, taking up various ancient speculations, the most recent being by Gassendi, I tried to ask myself how stones think. Nor do I consider these pages pure narrative play, and I attribute to them, if only privately, some philosophical dignity.

44Well, speaking of animal gnoseology, I would like to recall the cartoons of Tweety and Sylvester, where the canary, after glimpsing the cat, with a touch of perplexity says something like (I quote from memory) “I thought I saw a pussy cat!” Here, to explain how Tweety acts in a rather anthropomorphic way, I have to summon Peirce, and you’ll see that this immediately leads to the experience of hitting a wall, only I need it to happen in the dark. I am in a hotel room, I don’t know the place, I get up at night to go to the bathroom and two seconds later I feel a terrible pain. The pain itself is a primary experience; I think you feel pain and scream even before you realize if it is in your leg or hip. This is the phase that Peirce calls Firstness. Immediately afterwards I have the clear evidence that if I feel pain, it is because I bumped into a foreign body, and this is the area of Secondness, me and the world: something rises up and slowly defines itself in front of me. But we are still at the stage of the cry, or the stimulus-response process, the one where if the thing burns me I retract my hand, if it hits me I move over.

45Finally we come to what Peirce calls Thirdness. I tell myself that the pain is probably due to the fact that I bumped into the bedside table, or the chair, or the sofa... Now I know what happened to me, and I know how to behave accordingly. “I positively think I saw a pussy cat”. To go back to my Kant and the Platypus, I think there is fullness of perception, and perceptual judgment, only at this stage. First there are sensations, which are variously interpretable. (I apologize for the self-quotations, but it is my two interlocutors who keep bringing me up.) In 3.7.6 I speak of a driver who sees two bright circles coming towards him at night. I recall that a hare would look at them in fascination, or would move in such a clumsy way as to be run over, and this insofar as the sensations it receives are as “objective” as ours. We don’t die because beyond the firstness (very strong sensation of light) and secondness (something stands in front of me) we are able to move on to thirdness, on the basis of previous experiences: “it is a car that is coming towards me”, and from there we elaborate a series of movements to avoid being hit.

  • 16 Martin Mystère is a fictional character created by Alfredo Castelli for the homonymous comics serie (...)

46At this moment I am very uninterested to know if a dog would behave in the same way or not: if it is, I offer my congratulations, it has reached a very good level in the history of evolution. But, despite Ferraris’s good will, here we are not discussing the schema (in the subjective sense) of dog. Indeed, I think we owe an explanation to the readers: a topic that we would have liked to discuss, and maybe will discuss in the next issues, is the eighteenth-century and post-Cartesian polemic on the âme des bêtes; but it is clear that in this context the topic has not been taken up, so let us leave it alone. The question is whether this phase of Thirdness is independent of Semiosis. And I think it is completely irrelevant whether it is Semiosis that establishes Thirdness or the other way around (chicken/egg). The problem is that the moment of Thirdness, which for Peirce is that of the Symbolic, is that of Semiosis. If then this Semiosis is not verbal language is a secondary issue, and that’s why a dog could have Thirdness and therefore Semiosis, even if it can’t speak (for the most educated readers I refer to Java, Martin Mystère’s friend,16 who doesn’t know how to emit articulated vocal sounds, because he is a Neanderthal, yet he understands, thinks, writes on the computer, and is very intelligent), or a diminished Thirdness, like Sarah and other chimpanzees who have learned to express themselves by non-vocal signs but not to teach it to their offspring...

47The important question is that there is fullness of perception when there is fullness of Semiosis (or vice versa). When Ferraris says that the reference to language “excites us” because “it makes us too dependent on the spirit” he seems (contrary to his relations with the registry office) to express himself as a collaborator of Rinascita in the 1950s, for whom any reference to convention, culture, language (anything that was not production relations) sounded idealistic. But to say that the fullness of perception depends on our system of cultural assumptions (i.e. semiosis, or language, depending on the terminology) does not mean that our cultural assumptions are based on the multiplicity of intuition (at most they are its laborious anthology). It only means that they collaborate to give it an intersubjectively comprehensible and communicable form. Without Semiosis there is only Firstness and Secondness: I put my finger close to the fire, I feel it burn, I move it away and I scream. End of story. Many animals do this and we do not even know if they retain any memory of that unpleasant experience: moths seem not to, while dogs seem to – all the better for dogs, they are closer to Thirdness than moths, what a wonderful discovery. Nature is classist, it is the most politically incorrect thing there is.

48To say that the wealth of the multiplicity of intuition, to continue in our Kantian references, comes before the finalized perceptual judgment certainly means that perceptual judgment cannot claim that the multiplicity does not exist, not that the multiplicity of intuition replaces perceptual judgment. Between the multiplicity of intuition and full perceptual judgment there is something, you may not call it transcendental (and I don’t like it either), you may not call it Semiosis (call it Philip), but something must be there. That’s all there is to it.

49Otherwise we agree, in particular that the schema comes from an example, but we need to clarify one thing. One may well engage with Kantian philology, which leads to unavoidable mess, because in my opinion the way Kant speaks of schema in the first Critique is not the way he speaks of example and symbol in the third, and at most the concept of schematism is taken up (and transformed) by that of reflective judgment. But if this is what we want to do then I’ll throw in the towel, it’s not my job, I’m just a reader of Kant – from an early age, which means that, in terms of an ancient dichotomy of mine, I don’t intend to interpret him but to use him.

50So, let us try to see what can be understood today by the schema ‘dog’. Not something that comes before the experience of a dog (we have seen that not even Plato, let alone Kant, could have supported such an extreme form of Platonism, which would have required also the idea of platypus as pre-existing any trip to Australia). So let us say that, either from the experience of a Dachshund (or a Great Dane) or from the news I received about dachshunds (be careful Marconi, this point will also come in handy in the discussion with you), I create a drawing, a series of instructions, to recognize various types of dog, from the dachshund to the German shepherd, even managing not to confuse a Pekinese dog with an angora cat – which on second thought seems almost impossible, just imagine a Martian who sees them for the first time.

51This ability of ours to schematize not only on the basis of experience, but also through vicarious descriptions, is quite prodigious, but this prodigy encourages me to argue against Marconi’s objection (that I should somehow stick my nose into the black boxes of cognitivism). Of course, as a tourist I do, but as an author of travel books I do not. I don’t know how it happens that from the direct (ostensive) experience of a Doberman, from the verbal description of a Pekinese, from the drawn image of a Dachshund, we come to build a schema of ‘dog’ that allows us to recognize even a Labrador and a Greyhound. But this happens. My problem, as we will see, is to account for this semiosic process. What is certain is that we possess such cognitive abilities that we can overcome the silly philosophical querelle about whether we know the meaning of words by ostension or by definition (a problem that has occupied many unsuspected thinkers, from Augustine to Wittgenstein and Russell – a sign, as my grandmother used to say, that too much study makes you more of a dunce).

52We do not know the meaning of words by pure ostension. Let us imagine a child asking his mother what ‘dog’ means, in response to which she shows him a German shepherd. If the ostension remained such, the child would believe that German Shepherd is the name of that animal (Kripke’s syndrome) or – if the child had a minimal capacity for abstraction – he would be unable to recognize a Dachshund as a dog for the rest of his life. In fact, the child performs a series of very complex operations: he transforms the object into a sign, removing a series of properties that he considers marginal and fixing the attention on properties that he considers pertinent, from this selection he draws a model that could be valid for all the dogs in the world, and from that moment he applies it to other experiences, risking to call ‘dog’ also a cat or a wolf, except that the mother remains nearby and continuously corrects his schema, until it roughly adapts to the one shared by other people. The problem remains whether the child (Platonically) has a pre-schema of a dog, whether he has a tendency to schematize, to what extent previous experience affects his selection of the schematically relevant properties, etc.

53Does this training process take place through verbal means, vicarious images, ostensions or other semiosic means? I believe that different means can be identified for each subject. But in the end, the fact that people do not walk lions on a leash, or bring cats on a hunt, proves that there is some consensus. Except for statistically negligible margins, people agree on what should be understood as a dog – and indeed the Marconi/Ferraris debate between seeing a dog and seeing something as a dog seems, for the moment, secondary.

54So, the debate between abstraction from experience and transcendental schema is solved in this complex work that the child (or the urban savage, or those who learn a second language) carries out, without having an a priori schema of ‘dog’ – just as we do not have that of a platypus – in moving from incomplete ostensive experiences, descriptions of varying precision and inaccuracy, trial and error, successes and failures, rewards and punishments, to an ability to entertain meanings (the concept corresponding to the word ‘dog’ as different from the concept corresponding to the word ‘cat’). And please, Maurizio, do not try to insinuate that in this process a native aesthesic experience would prevail over language. I would not believe it even if I were a dog, let alone a professor.

  • 17 The sentinel, which in Italian is a female noun (translator’s note).

55I like that Ferraris likes the criticism I make, at a certain point, of a confusion between grammatical and semantic categories. It’s true, no Italian saying “LA sentinella”17 would thereby think that the soldier in question has female hormones. But my wife who is German tells me that, after thirty-five years in Italy, she still thinks of the sun as a generous mother and the moon as a cold gentleman, or something like that. So, without throwing away the usual baby with the usual bathwater: interpretation is always an interpretation of something, there is a thick peel of multiple intuition that determines the course of our perceptual judgments, but this does not mean that there is an auroral and fresh primality of sensation that precedes, very clearly, every cultural and linguistic knowledge. Bring me a hare that doesn’t get run over by a car with the headlights on and then we can talk about it.

56And now let us move on to Marconi. I believe (based on what I have written in the previous paragraphs) that I can agree both with Ferraris that we reach schemas starting from examples, and with Marconi that it is not a matter of images but of procedures (after the long work I described in the case of the child who is shown a dog). If good taste and modesty did not prevent me from doing so, I would refer the reader to the second part of my Theory, the one dedicated to the modes of sign production, in which I discuss at length the difference between congruences, projections and graphs. I was writing at a time when I was more convinced than now that everything could look like something else under a certain description, so of course (although my sources were probably different) I found it problematic that an image could correspond to an object. At that time I believed that there was a group of naive kids to fight against, i.e. Thomas, the first Wittgenstein and the Lenin of Materialism and Empirio-criticism – all people who believed in picture theory, Abbildung and adaequatio. But even in my anti-iconic fury I could not deny that there were mapping procedures (like the portula optica of Renaissance painters), so one could always tell if something was not an image of something else, as much as one wished it to be. This was possible precisely through mapping procedures, connecting a presumed Mapperx to a presumed Mappedx.

57In the light of these arguments (which I take up again in Chapter 6 of Platypus) I also believe that the difference between seeing and seeing as can be traced back to the distinction (which I make there) between Alpha mode and Beta mode: I would say that both the hare and I see two light sources on the road at night, except that I see them as a car. But this is not the point I want to discuss. The point is, my relationship with Marconi risks becoming difficult, because for party reasons he deals with (analytic) Philosophy of Language, whereas I deal with Semiotics, and everyone thinks that we both deal with the same thing, except that for analytics semiotics are all “speculative” know-it-alls, and for semiotics analytics are people who do not want to work and keep rewriting the same four canonical texts. Fortunately, Marconi is a heretic for the philosophy of language as I hope to be a heretic for semiotics, but this does not detract from the fact that the two of us deal with two absolutely incomparable things, even if (judging by the examples and the bibliography) it seems that we have many points in common.

58As a heretic of the analytic philosophy of language, Marconi got it into his head to understand how it happens that we think and know things. He is so interested in this problem that he decides, in his book, to replace the notion of meaning (something fundamentally unreachable) with that of two sets of competences, the inferential and the referential one. The difference is not so much that the nature of these two competences appears new in the landscape of semantic discussions: if the inferential competence is the one that allows us to say that cats are animals, and the referential competence is the one that allows us to recognise a cat according to some of its properties, this distinction simply reproduces the one between Dictionary and Encyclopaedia.

59Rather, Marconi insists on talking about competences because, for example, inferential competence is a skill that allows one to navigate between lexical networks, and referential competence is a combination of knowledge and procedures, but neither of the two can be reduced to an entity called meaning, or to the knowledge of such entity. In this sense Marconi goes beyond the old debate between dictionary semantics and encyclopaedia semantics, because in that perspective dictionary indexes or (recorded or recordable) encyclopaedic properties are components of meaning, while for Marconi knowing that a dog is an animal only means being able to navigate in the hypertext of socially established uses by choosing the right links, i.e. the most frequently used ones.

60On the contrary, I am much more old-fashioned. All my life I have continued to believe that we can talk about Content (and terms and utterances). I am a speculative philosopher, not an analytic. I fake hypotheses, not like that foolish Newton who said he didn’t, and then even believed in occult forces. I am interested in the question “what corresponds, somewhere, to the content of the expression dog”? Since there is still no camera to tell us what corresponds to it in the mind of the speaker, I can only resort (with Peirce) to chains of interpreters: I record that people tell me that by dog they mean this and that, and they provide me with drawings (even bad ones) and sounds (like “woof woof”) and other signs (even a hypothetical vial containing a dog’s smell, because the smell of a dog is different from that of a hen – we haven’t been to the countryside for a long time, but I swear it’s true, people talk and think even using information of this kind).

61Then it is clear that, even if in my last book I say that it is no longer appropriate to judge every mentalist discourse or research as taboo, I am a child of a philosophical era in which mentalism was worse than paedophilia, and therefore even when I do not notice it I am a repulsive behaviourist. To know what goes on in people’s minds when they talk about dogs, I have to induce them to produce expressions, and other expressions that interpret the former, and so on until I can without losing my mind. And so what I call Cognitive Types are pure hypotheses, just to account for the fact that Marconi, Ferraris and I do not quarrel when we say that Tom is a cat and Jerry is a mouse (I am of the opinion that something in our head works the same way). But to know what the hell of a cognitive type one has in mind I can only know through the interpreters they give me, whose statistical average gives me what I call Nuclear Content, a definable cultural object, albeit accordion-like and with frayed edges.

62If this is the case, I do not find it astonishing that the cognitive type (and the nuclear content) should be made up of the results of different abilities. Let us imagine that we are interested, like Dewey in Art and Experience, in defining the nature of an accomplished experience, that is, the phenomenon that Dewey called fulfillment. As an example of a perfectly accomplished experience Dewey chose a dinner in a restaurant in Paris. Very American, very 1930s, but in short, if we accept it as a valid example, that experience is a whole resulting from different cognitive structures and sensory phenomena (taste, touch, perfumes, ability to assess the waiters, the quality of the silverware, the very way of tasting the wines, etc.). What worries us is not at all that such different skills and processes collaborate to the fullness of experience: the question is whether we can speak of a single, organic experience – in the same sense in which we can define that entity which is an opera, even if different arts converge on it, from instrumental music, to singing, dancing, mimicry, acting, scenography, lighting, etc. So I don’t find it disturbing that the cognitive type of a mouse is comprised of inferences, images, frames, timic connotations etc. Just as Marconi heretically tries to throw away meaning (once again – heretics always renew themselves in their mad bestiality) by dissolving it in a network of competences, I try to throw away the idea that meaning or content can be defined in homogeneous terms, because it is made of different experiences and notions, resulting from different cognitive structures.

63Essentially, I think that Marconi redirects me to the black box (where I should stick my nose) because it seems right to him that I should say something about all those cognitive processes that I summon on the scene. Which would be true if I were interested in defining all those cognitive processes. But I am only interested in saying that, if we want to talk about Content, we have to think about a cauldron that records the results of different cognitive processes (and of course expresses them for interpreters). I am not saying at all that the Cognitive Type is a single cognitive structure. I’m saying it is the more or less standardised package of experiences that we pull out of our pockets when we have to understand a given word, or recognise the objects to which this word may refer. It is a package, a survival kit, a Swiss Army knife, made of different things, but which works as a unitary tool when needed. I repeat: I don’t know how it works, I don’t even know where it is, I just know that if I ask my subjects what they mean by dog, cat and the like, little by little, even if I have to pull it out of them, they provide me with a series of interpreters, images, notions, smells, sounds... Marconi suspects that this is the hypostasis of a set of cognitive functions, but by saying so he still thinks in terms of those who are interested in knowing how we think. I say it is the repository of knowledge, because I am not interested in how we think, but what we think about.

64Let us take an example. I just came back from an island in the Mediterranean where, one day, I got stung in the arm. I thought it was a jellyfish, but the local experts told me otherwise (in fact I experienced how jellyfish sting in the following days, and it’s worse). They told me that my relatively minor injuries were due to the sea spaghetti, which are found right near the rocks. Now, I think I have a fairly articulate cognitive type for jellyfish, I’ve seen them alive, I’ve seen them in books, I know how to look underwater to spot them (better if against the light), and I could express by various means what I know so as to provide the Nuclear Content corresponding to my Cognitive Type. I repeat that Cognitive Type and Nuclear Content are shared by average speakers, whereas more in-depth forms of knowledge belong to the Molar Content. Days later I came across an unusual animal, which I could describe as a large sauteed egg, the colour of melted-butter, stirring over a half-cooked Idaho potato, on a background of dwarf cauliflowers. I realised it was a fearsome beast but I wasn’t sure it was a jellyfish. I caught it with a bucket and a superexpert then told me that it was a terrible African jellyfish, which had arrived on those shores due to some mysterious play of the currents. This is to say that our knowledge about the content of the term ‘jellyfish’ can always widen, but for everyday needs (in a temperate zone) what I knew before I met the egg in the pan was enough.

65Instead, I didn’t know anything about sea spaghetti (I guess it’s a local name). The first indisputable fact was that it was a stinging agent. You could say that this was a nucleus of inferential competence, received by cultural transmission; but in this regard I was uncertain whether they were vegetable or animal, as happens for many of the things that are lodged between the rocks. I proposed to look underwater to discover them, but the only instruction to identify the referent was expressed through a metaphor (“spaghetti”), from which I could extrapolate the instruction “look for threadlike and possibly white objects”. Mind you, the metaphor could have referred to uncooked spaghetti, and then I would have had to look for rigid threadlike objects. In fact, I looked for soft, and therefore fluctuating, objects. I wonder why, but I think it depends on a tacit metaphorical rule, that the metaphorizing object must be taken in its most usual state, at least for us (if I say of a maiden that she is a fawn, I mean a living, moving fawn, as usually seen by the hunter, not asleep, embalmed or ill).

66I looked around and saw many filiform vegetables, waving like Hawaiian dancers, of sloping colours from whitish to yellow and green, and I could not understand which were the fearsome spaghetti. However, for days to come I avoided putting my feet on every spaghetti-like substance. Well, I am an example of a speaker who has very poor knowledge of the meaning of the term ‘sea spaghetti’; I have difficulty identifying the referent, but I have developed a certain ability to identify a class of possible referents, and on this basis I could give instructions to someone who asked me how to avoid them. I cannot explain what they are in terms of inferential competence (in terms of Molar Content I am fully incompetent and would fail a natural science exam). However, I would need five minutes of conversation with the local expert, or ten minutes of encyclopaedic consultation, and I could move with a certain ease, both as a user and as a master of language.

67How many cognitive functions have been put into play for such a simple “game”? I’m afraid to say. My problem is not to identify them: it is to solicit interpretations, one after the other. Maybe some could draw sea spaghetti with impressive realism, others couldn’t, but that doesn’t mean that the former have a Cognitive Type and the latter don’t; simply, the margins of the two CTs are different, but in the end if they know what that word means, both sides should be able to tell me how to recognise the object, and those who can’t draw could come up with perhaps very poetic metaphors, like “green things dancing on the rock like Salome under the last veil” – on the other hand, isn’t it precisely what whoever invented the expression “sea spaghetti” did?

68I believe that Marconi in this phase of his work is very curious to know what and how many these cognitive procedures are. I am a little less curious – it is enough for me to know that there are many, to see them all playing together, and to dismiss the structuralist lexicographers or the post-Chomsky dictionary lexicographers who continue to play with Human Male and Adult, believing they are solving the mysteries of linguistic usage.

69I think it is important to understand this difference in focus, perhaps in vocation, because it is only by understanding it that some apparently controversial points can be resolved. But we will have to talk about this in the next episodes, when, as Ferraris says, we will have to discuss not how we schematize dogs, but how dogs schematize us.

Torna su


Eco, U. 2000, Kant and the Platypus: Essays on Language and Cognition, tr. by A. McEwen, New York, Harcourt, Brace&Co [Kindle Edition].

Ferraris, M. 1997, Estetica razionale, Milano, Cortina.

Marconi, D. 1997, Lexical Competence, Cambridge (MA), The Mit Press.

Paternoster, A., Meini, C. 1996, Understanding Language through Vision, “Artificial Intelligence Review”, 10.

Torna su


1 Or of the underlying concepts. Here the distinction is unimportant.

2 Marconi 1997: 79-82, 145-149.

3 See e.g. Paternoster, Meini 1996: 1-2, 37-48.

4 Ferraris 1997, § 3.6: 344.

5 Eco 2000: 123.

6 Cf. Eco 2000: 131.

7 Cf. Eco 2000: 155.

8 In my book Lexical Competence, I try to show that the neuropsychological data suggest that application (including to abilty to generate pictures) and naming (including the ability to identify pictures) are different abilities, even though they share some data structures (see Marconi 1997: 69-70).

9 Eco 2000: 168.

10 Ibidem: 163.

11 See Marconi 1997: 68.

12 Italian saying meaning “attempting the impossible” (translator’s note).

13 I would like to provide a few supporting elements. Apparently, there is nothing more distant from transcendental philosophy than the tabula rasa argument. But we can explain the whole transcendental philosophy precisely through the model of the tabula. The concepts of the intellect «spring pure and unmixed, out of the intellect, as absolute unity» (B 92 / A 67). One cannot rule this out, much like the idea that the tabula is not part of experience. It would be much more problematic, however, to think that these principles are activated without external solicitation (if this were the case, Locke’s objection to Descartes about children and idiots would be very pertinent). Kant openly argues that forms of knowledge (both sensible and intellectual) are put into practice «on occasion of the sense-impressions» (B 118-19, A 86-87). Kant has long doubted the possibility of synthetic a priori judgments, and in the 1770s (cf. Reflexionen, fns. 290 and 500) he observed that all analytical judgments are rational, while all synthetic judgments are empirical. The point can be this: is empirical the same as a posteriori? Can there not be an empirical apriori? This is perhaps exactly what empiricists (and not only) had considered under the title of tabula rasa.
The thesis according to which the transcendental is not every a priori knowledge, but only that through which we know the «what and how» of intuitions or concepts («representations» in the sense of «re-presentations») are possible only apriori (R 80-81 / A 56-57) is not so much an act whereby the transcendental represents a sphere within the apriori, but rather the sign of a non-coincidence whereby there may be a transcendental that does not exactly coincide with the apriori (which is partially the case with the tabula). The transcendental logic is chained to the empirical from two points of view: 1. it is subject to time, time to the trace, the trace to space; 2. it presupposes that somehow knowledge is already always initiated, if it is true that the condition of this pure knowledge presupposes that objects are given (B 87 / A 62). That there is a difference between the image as an object and the monogram as an exemplary function is quite straightforward. That this rather resembles a method, i.e. a system of traces, is also entirely plausible. That it should precede experience is conversely much more problematic. After all, a posteriori means for Kant that the object makes the representation possible (B 124-5, B 92); and so there is no knowledge that is not empirical (according to Kant’s fundamental assumption). That the representation makes the object possible, according to the character of the a posteriori described in the same passage, is even more incomprehensible. In what sense can a representation make an object possible? At most, a re-presentation can make it possible, but this is equivalent to saying that the transcendental is the tabula. And this is what Kant means, after all, when he maintains (B 132) that the representation of pure apperception is the I think, that is, that cogito which was traditionally believed to have no image, and which arises only with some image of external experience or thought content. In fact, the representation in question is nothing but what ensures the constant unity of self-consciousness (B 135), i.e. that all the experiences and thoughts that concern me are in my tabula, and not, for example, in someone else’s. The crux of the matter is set when Kant says that all knowledge necessarily occurs through experience, and when, in the chapter on schematism, he gives the example of the schema of the empirical concept of dog. The synthesis, we read in § 24 of B, is transcendental and purely intellectual, defining the unity of apperception. Is this different from saying that the tabula must be presupposed as prior to any experience, and that therefore we have no experience of the tabula? It is an a priori form. Here too, does this clash with the description in terms of tabula? Indeed, we have a faculty (capability, capacity) to believe, just as we have a field of vision, that it seems to us (perhaps) to be round, but it is not, and is rather amorphous. To say that the tabula, or the eye, are a priori does not in any way hinder what has been said so far. The problem is rather that Kant must maintain that the imagination is spontaneous in its productive function. But if it were truly spontaneous it should produce, for example, a purely imaginary world; why does it connect to this very world? Simply because it performs a passive and reproductive function – retaining the sensations, or even thoughts, although Kant argues that intellectual synthesis does not need imagination. Yet it is clear that it needs memory (after all, elsewhere, B 103, A 78, Kant had argued that synthesis in general is the result of imagination) – which must be assumed to be a priori, and therefore does not happen after experience (as perhaps Hume would have said), but with experience, and, more precisely, an instant before it (since otherwise experience would not be possible; but this is only like saying that the target pre-exists the arrow). After all, Kant makes many partially contradictory statements. For example, he writes that the same function that unifies representations in a judgement gives unity to representations in an intuition, and is a pure concept of the intellect (B 104-105, A, 79), so that synthesis is sometimes attributed to the imagination, and at other times to the intellect (and obviously to the senses, we would say following the Aristotelian koinè aisthesis). The point is rather that for Kant the categories are inferred from judgements (i.e. from syntheses), and that this happens openly and not in a rhapsodic form. This is precisely what seems difficult to grant to Kant, because, when he said he was describing the experience of us humans or of beings similar to us, he had already determined the aposteriority of the a priori (or rather, a certain relative apriority). Could we say that a blind person lacks that a priori function of vision which for a sighted person is, for example, the retina? No, they simply lack a sense organ. Kant’s affirmation of not having drawn any of the categories from elements of sensibility, unlike Aristotle (B 105, A 79-80) is therefore to be understood simply in the sense (think for example of Theaetetus) that it is not with the senses that we say that something exists or not.

14 Eco 2000: 52.

15 Cf. Eco 2000: 24.

16 Martin Mystère is a fictional character created by Alfredo Castelli for the homonymous comics series published in Italy since 1982 by the publisher Sergio Bonelli (editor’s note).

17 The sentinel, which in Italian is a female noun (translator’s note).

Torna su

Per citare questo articolo

Notizia bibliografica

Umberto Eco, Maurizio Ferraris e Diego Marconi, «The Dog Schema»Rivista di estetica, 76 | 2021, 10-39.

Notizia bibliografica digitale

Umberto Eco, Maurizio Ferraris e Diego Marconi, «The Dog Schema»Rivista di estetica [Online], 76 | 2021, online dal 01 mai 2021, consultato il 24 juin 2024. URL:; DOI:

Torna su


Umberto Eco

Maurizio Ferraris

Articoli dello stesso autore

Diego Marconi

Articoli dello stesso autore

Torna su

Diritti d’autore


Solamente il testo è utilizzabile con licenza CC BY-NC-ND 4.0. Salvo diversa indicazione, per tutti agli altri elementi (illustrazioni, allegati importati) la copia non è autorizzata ("Tutti i diritti riservati").

Torna su
Cerca su OpenEdition Search

Sarai reindirizzato su OpenEdition Search