Frogs, Pigeons and the Synthetic A Priori Within Us All: Can McCulloch's Search for Universals Tell Us Anything Useful in Perception Debate within Philosophy of Art?

Marko Kardum

Contemporary aesthetics, or even better philosophy of art, considers the ontological status of artworks to be somewhat problematic, especially with radical examples such as ready-mades that were highlighted by some philosophers of art, namely Arthur Danto. These radical examples are based in their perceptual indiscernibility – one cannot differentiate everyday objects from artworks. Therefore, the only property of such objects that can turn them into artworks cannot be formal or perceptual. What proves to be important for such a claim is human perception. With trying to prove his insights in philosophy of art, Danto concludes that vision has no history and to support that claim he follows Fodor's model of modularity of mind. However, there is a sense in which we can say that vision has a history and that can be found in development of so-called schemas which can replace what we know as universals. Search for these universals was one of the most recognizable features of Warren McCulloch's work and it can be traced in several of his papers. This paper tries to show that McCulloch's work would be a much more solid foundation for analyzing perception even in terms of philosophy of art in which, unfortunately, McCulloch's name never appeared.


Let me start this chapter with a short explanation of the title and a short personal backstory. Even as a student I was interested in the field of philosophical aesthetics (a field I still work in quite a lot and which is still of great interest to me). Thanks to some well-known authors in this field I became even more interested in some of the problems of contemporary aesthetics, if I may name it such, and their relationship to even broader problems of analytic philosophy. During the preparations for writing my doctoral thesis, I became increasingly concerned with the problems of ontology of art, specifically with those that pointed to the problematic nature of the very essence of an artwork. These problems indicated there were some serious issues in defining an artwork, especially from the avant-garde onwards, and it wasn't hard to comprehend why this was the case. Avant-garde art practice showed that literally any common and everyday object could become a work of art. One question that naturally arose was: what was it that could so dramatically change the ontological status of an object, changing what once was only a common object into a work of art? One of the possible answers to this question was given by a well-known philosopher of art, Arthur C. Danto, especially in his influential book The Transfiguration of the Commonplace: A Philosophy of Art [1]. So my work became even more focused on his philosophical insights. His view of the aforementioned problem was as elegant as could be: there is no difference whatsoever in physical properties of some common everyday object and his counterpart we consider to be an artwork. What makes it possible to discern them is a property that is hidden from our eyes, it is invisible and belongs to the history of art and art's relation to the context in which some object appeared. If it is possible to determine such a relation between an object and its historical context, we are dealing with a work of art, and we can interpret it and determine its value. Bad art is still art. If such a relation cannot be determined at all, the object in question is just a common everyday object. In a given case no interpretation or evaluation typical for artworks makes any sense as we usually do not interpret our forks, cars, or light bulbs as art. Of course, there are some relations that are visible to us, for example we can see one thing being in front or to the left of another, but these relational properties obviously are not of that kind. The said extension, namely that from the physically perceptual to the relational properties of an object, proves to be a cornerstone for Danto's theory of art and its fundamental problem of ontology of an artwork and is sometimes referred to as a specific kind of institutionalism or institutional definition of art. Being more or less plausible to other philosophers of art and art historians, Danto's solution necessarily brought up two more issues that are very important for contemporary aesthetics: 1) it became increasingly difficult to claim that the problem of ontology of art still belongs to philosophical aesthetics so one should rather think of philosophy of art when dealing with given problem – a thing that shall not concern us much at this point – and 2) since what allows us to distinguish works of art from those objects that fail to make it into the same ontological category is invisible to our eyes, problems such as the ones concerning human perception, should not be relevant anymore. However, human perception proves to be one of the crucial concepts one has to plausibly explain in order to successfully approve or disapprove the extensional definition. And, actually, by no means should that be strange. If human perception is fixed once and for all, then all cultures and each individual within them can only see any object of perception in the precisely same way (I am disregarding specific angles or light sources that are already “calculated“ in this claim and are not quite the subject of our interest in this case). So symbolic representation, interpretation, and higher-level perception which includes interpretation is not a question of human biology and physiology of our brains changing over time in interaction with our visual practice. According to Danto, our visual perception is given and fixed or hardwired and it should be sharply separated from higher-levelled processes such as symbolic representation and interpretation. This way, our way of seeing is always the same, and the described relations specific to art belong only to social and historical context. This is why he is able to say that not every work of art is possible in every time period. Naturally, Danto's critics (here only regarding this specific question) consider this model of perception to be false. They tend to show that our visual practice influences the way we see at a fundamental level. In other words, the eye has its own history and although it is possible to repeat Danto's words about some works of art not being possible in every time period, the motivation for it completely changes. For them it is not the case that our interpretations and representations would not work in specific historical context. It is about human beings not being able to see in the same way in different periods of time, and precisely because of the practices of showing which can change the way we see. This way, what could have been considered as moving away from aesthetics, at least in terms of discussing beauty or taste, actually returned on a level much closer to the origins of the term aesthetics – a very basic notion of senses and perception, and a subject that is always interesting to those who wonder how to transfer processes that occur in animals', including humans', brains into artificial entities; a legitimate task for any computational enthusiast.

What Danto did to defend his own ideas was to introduce some cases of animal perception. He wanted to show that at the lowest level of our perceptual abilities lies the same mechanism which can be found anywhere in nature. Namely, he used pigeons and their visual perception as an example with which he sought to prove his thesis of “impenetrable“ visual perception. He addresses this in an article known as The Pigeon within Us All: A Reply to Three Critics [6]. Of course, Danto was not the only one to use animals' perceptual mechanism as an example. While writing my doctoral dissertation I was unfortunately completely unaware of Warren S. McCulloch's work and one of his seminal papers What the Frog's Eye Tells the Frog's Brain [7]. This paper showed some unexpected, yet quite interesting results that could be taken as a steady ground for further research. As it turned out, that paper should have been treated as an empirical extension of McCulloch's and Walter Pitts' paper named How We Know Universals: The Perception of Auditory and Visual Forms [8] and both of those papers would eventually become very interesting to me regarding questions I was seeking to answer. But never would I have even dreamt that this could be a way of dealing with those questions. Cybernetics and computing theories were not something that stood in an obvious correlation with philosophical aesthetics, let alone philosophy of art. The story of ending up writing about McCulloch and his work is a completely different and irrelevant one. Yet, frogs going well with pigeons, at least in a title, could be one of the motives. The other is a specific approach that McCulloch was taking in his works. And that approach is always looking for solutions well hidden beneath the surface. In his case it is the old residue of Kant's understanding of general conditions for acquiring any knowledge stated in his term of a priori, but shaped in a new way that met the standards of questions being current in the 20th and 21st century. And maybe not just current that could actually be an understatement depending on which corner one sees it from – but also crucial. It is probably possible to look at McCulloch's work as philosophy or metaphysics but I seriously doubt it can be disregarded as purely theoretical or uninformative about the world we live in. Unfortunately enough, the philosophical value of his work somehow often falls into a pit and it seems that, like its influence, it is far more recognizable and appreciated in computer sciences or cybernetics than in philosophy, let alone aesthetics or philosophy of art. This paper is just a clumsy attempt at bringing the value of McCulloch's work for philosophy out in the open. Whether it makes any sense or whether I am just seeing connections where there are none is, of course, open for discussion.

All that being said, connection between some problems of animals' (again, humans' included) perception that unexpectedly rose from philosophy of art, conditions for perception per se is formulated and expressed in Kant's term of the synthetic a priori, and the possibility of transferring it to artificial entities is something I will try to address more closely by presenting this paper. For now it was enough to roughly sketch the idea behind the paper's title and the problem of ontology of artworks as it appears in philosophy of art.

Is a shovel really only a shovel? Or shall we call a spade a spade?

What we have learned from the avant-garde artists and some philosophers of art is that everyday objects contain something more than just formal properties that meet the eye and that they can become artworks. Or at least we have been told so. Consider here for a moment the work of famous artist Marcel Duchamp and his well-known “sculpture“ called In Advance of the Broken Arm that actually, in its most basic form, is nothing more than a shovel. Duchamp used to do similar things with many everyday objects, he would have simply designated them as artworks thus proclaiming their new ontological status. Warhol's Brillo Box is another example of this newly established art practice. Obviously, the new paradigm of producing artworks in turn caused significant problems and collapsed all of the standard definitions based on perceptual formal properties of any imaginable artwork up to that point. What Arthur Danto came up with was a different approach to artworks. According to him, any object could become a work of art if it met following criteria: 1) that object is representing something or is simply about something (it could be even about itself and its own place in the history of art), 2) it is about something in a specific way or has a particular style and 3) it establishes relations in an historical and social context. These conditions formed the necessary and jointly sufficient reasons for an object to be considered a work of art. Thus the formal and noticeable properties of objects became irrelevant for defining artworks and, ontologically speaking, shovels are possibly not only shovels. In his further elaboration Danto describes even more indiscernible pairs of objects with one being just an everyday object and the other being a work of art. But his solution did not come without a cost. Danto was now obligated to consider some implications this solution had for the way we understand human perception and the way the human brain actually works.

To defend his own position, and by extension that any artwork can only be determined as such by interpretation (which has nothing to do with our perceptual mechanism and visual inputs), Danto argues that human vision does not change. Humans have only been able to see in one way regardless which culture or historical period individuals belong to. For him, the human visual mechanism is hardwired, and the claim that our way of seeing reflects changes in our (artistic) ways of representing the world gives “a far greater plasticity to our optical system than the facts of perception seem to allow“ [9, p. 1]. So any idea of so-called historicity of the eye should be rejected. Furthermore, he tries to show that at low-level perception activity, and that counts for just gathering visual information from our environment, we as humans do not differ from pigeons [6]. This position is further elaborated with some of the insights regarding pigeons' behavior when stimulated by the objects that are recognizable to them. In short, when faced with photographs of different objects, pigeons can recognize the basic shapes and forms of those objects that their natural environments are built of, such as trees for example. That such recognition really happens is obvious from pigeons’ behavior – in order to get food, they simply peck one button when the picture is recognizable and another when it is not. To prove his point and explain why pigeons react in this way Danto reaches for the ancient theory of eidolon, which basically consists of an object and its “phantoms“ that are intercepted by the eye and that, in turn, enable us all to see and recognize that particular object. The same process of objects emitting phantoms that our eyes catch takes place regardless of any other contingent circumstances such as cultural affiliation, beliefs, knowledge, past experiences etc. Therefore, there is a small pigeon somewhere within us all that simply gathers visual inputs and sends this information further up our neural structure until they reach our higher-level center. It is where we can start to differentiate conventions that we agreed upon in representing objects of the world and visual information that have actually been given to us. Conventions are not built into our visual system. From this point on Danto allows a switch which enables us to interpret what once was only a shovel as a work of art:

At a level higher than that of optical reality, there is no doubt that people see the same things differently at different cultural moments […] but a robust theory of the eye as historical would require that whatever accounts for these differences penetrates the optical system in such a way that the eye itself changes with history so that, at the level of ophthalmology, individuals see the world differently, or even, in the strongest version of the thesis, see different worlds. [9, p. 1]

Here it is possible to see all the work needed for interpreting an everyday object is done. Who would possibly think the human eye itself could change? When I look up at the Moon am I not seeing the same thing as an ancient Greek did when he looked up at the Moon? There is nothing that can penetrate our optical system in the described way and that is more than obvious. And in the term “penetration of the (human) optical system“ lies what Danto was preparing his ground for. Here is what follows as a conclusion in Seeing and Showing:

Visual processes are […] cognitively impenetrable. How we see at the basic level relevant to adaptation is unaffected by what we know or what we believe, a s much so as cell division. The visual system is, to use the deep conception of Jerry Fodor, modular. Modularity means a segregated default system, which functions in independence of other systems. Lucky for us that it does! [9, p. 8]

So our perception is modular, with low-level processes being out of reach of our cognition. By later on stating that “we almost certainly share the main features of our visual input system“ [9, p. 9] Danto concludes that what really is historical in our perception is our central system which enables us to understand meanings which are not meant to be seen simply because, unlike animals, we are historical beings after all. This way history supervenes on perception. Of course, things do get a bit more complicated than that. And after all, is the central system really historic as he understands it? Or should we just say something else in it changes historically, with the system itself at the basic level being just the same throughout history? I see no reason to consider it being different in its fundamental physiological level, just as the eye. And if interpretations that are governed by general rules and propositions can change in our central system, then they should be able to change even if located outside of it. And it seems this is the case. Semantical or symbolical propositions can be found outside the central system. And that goes against Fodor's syntactically only defined subsystems or modules.

The “history of vision“ debate took several forms and one of those forms can be reduced to the question of penetrability of our visual system by our central processes or to the question of our overall perceptual mechanism. To me, it seems that this particular form of debate does not have a clear winner yet, and I would not even dare to claim it will soon have one, although I will later take a side, so to speak. Besides the question whether this kind of modularity can be defended, there are attempts to explain the “history of vision“ debate as a debate over sensory stimulation and over history of our attention [12, pp. 136-159]. Let us here briefly explain why this paper focuses on modularity of perception as it was earlier explained by referring to other stated questions. It seems that sensory stimulation can hardly be mistaken for something that does have its own history – retinal processing of stimuli has not changed, and one can expect it will stay this way for as long as we talk about humans. History of our attention focuses on our experience of objects, and as such can be regarded as perceptual phenomenology. This, however, is for sure part of our interpretation process that Danto and other representatives of the said modularity would not try to disprove at all. It is not the question whether stimulation is physical or whether interpretations that can vary for several reasons can change our experience of an object. Therefore, what we should consider is the way modularity was used by Danto, or even more precisely, the way it was set up by Fodor. And this is how we can establish a connection between the perception debate that took place in contemporary aesthetics, namely philosophy of art, and McCulloch's work.

In his well-known monograph The Modularity of Mind [14] which Danto also referred to, Fodor argues for a computational theory of mind that is set up syntactically and in which the brain is divided into modules or subsystems, such as the perceptual subsystem. Those subsystems or modules are 1) domain-specific, 2) innately specified, 3) associated with distinct neural structures and 4) computationally autonomous. Therefore, the perceptual module is computationally autonomous and informationally encapsulated, [15, pp. 335-336] which actually means precisely what we have already read in Danto's work – our perception is impenetrable by our central system and cognitive powers. There is no feedback, external or internal, to our innately specified mechanisms adjusted to serve specific purposes whatsoever. Modules are independent and perceptual inputs are in no way affected by its outputs. Moreover, our understanding of the central system must differ significantly from understanding of modules or subsystems.

Are computationally autonomous and informationally encapsulated features of perceptual modul together with the theory of eidolon really the way to explain our perception? It seems that in the case of eidolon theory we are faced with an obsolete theory which can even be described as a kind of naive naturalism. Even more so, it highlights only what some object can do, disregarding subjects' power to intervene in the process and that goes against Kant's notion of an active subject that is modeling the world. Granted, Danto's use of eidolon seems more as an illustration than a reliable theory on which we can build our understanding of perceptual processes. But even as an illustration it depends heavily on modularity as explained by Fodor. And there is more than one problem regarding that explanation. I will address those problems later, especially in light of McCulloch's work.

McCulloch's galaxy

There are lots of anecdotal stories about McCulloch that often describe him as an interesting character but which also shed a very specific light on his theoretical and experimental work, together with the relationships he had with other members of his famous group of associates such as Walter Pitts, Jerome Lettvin, or John von Neumann just to name a few, and I would here like to recommend a wonderful book edited by Anderson and Rosenfeld [18] that got me really interested in McCulloch and his colleagues. In addition, one of his papers titled What Is a Number, that a Man May Know It, and a Man, that He May Know a Number? [19] which could possibly point to a thin line between philosophy, neuropsychology, and cybernetics – a specific feature of McCulloch's work.

There are at least two of McCulloch's papers that I find relevant to the subject of this paper. To be fair, I am sure there are more of them but here I will concentrate just on the two already mentioned papers: How We Know Universals: The Perception of Auditory and Visual Forms, and What the Frog's Eye Tells the Frog's Brain. In a sense, the latter can be considered as an experimental and empirical extension of the former so they will be considered in that order. In How We Know Universals McCulloch (and Pitts), by studying pattern recognition, showed three important improvements in building networks: 1) brain is structured and layered, 2) perception is more subtle than we think and 3) visual inputs can control motor outputs through distributed activity of a layered network without executive control, showing example of “cooperative computation“ [20]. So in this way, perception should not be considered only in terms of top-down or bottom-down theories, but in terms of layered detectors that search for certain features of objects of perception. In recognizing patterns we have to ask ourselves how we know what those patterns can form. That is the question of universals. Basically, it is a philosophical problem of being able to subsume any individual object under the universal category. Differently put, it is about how we think of any possible triangle that we see, red or a yellow, equilateral, left- or right-angled, as a triangle. According to the mentioned paper, we can do it in two ways: 1) by gazing to center on the middle of a visual pattern, which in turn shows that different points of retina activate the most appropriate muscles to “hold“ the object in the center of our visual field and that optical subsystem can work without the central processing of visual inputs controlling motor outputs, and 2) by extracting features of the image and forming an average that is always relevant in relation to only one universal. The idea behind the first approach is that we should not treat the brain as a firm hierarchy but more as cooperation of subsystems that connect sensations and behavior. This is something that goes against the Fodorian model of informational encapsulation and computational autonomy of modules or subsystems. McCulloch's and Pitt's model showed that subsystems are strongly connected and presented perception as an interaction of various subsystems rather than a process that takes place in only one subsystem that sends information further up to central processing where it is possible to interpret or to recognize representations.

So far we have seen the role of pigeons, and now it is a good time to introduce frogs. In What the Frog's Eye Tells the Frog's Brain by experimentally and empirically following insights from How We Know Universals, primarily Lettvin and Maturana developed McCulloch's ideas through psychology and anatomy. But there is still a strong resonance of McCulloch's interest in universals. What the paper showed was fairly controversial at the time, mostly because the results presented didn't match the usual expectations for the mammalian cortex. What they showed is that frogs have different detectors whose outputs were sent to different regions, thus confirming the layered structure of the brain with distributed activity for each action that some organism may conduct. Of course, different organisms conduct different actions related to specific tasks which are important to them, so retina could not be understood as a general purpose device anymore but rather as a specific one due to natural differences between organisms and their general abilities. These results transformed our understanding of the brain and proved that there can be no modularity in the Fodorian sense. However, what both papers had in common was McCulloch's search for the way universals work. And it is here where Kant and his term a priori fit well with the animals which appear in the title of this paper. Again as in Danto's case of eidolon, it is granted that results presented apply only to frogs, but it is precisely this search for a priori that can be generalized and applied to any brain.

Putting aside different interpretations of Kant's philosophy, especially those concerning his famous concept of transcendental idealism [see 21], his notion of a priori will be presented in the shortest possible way. After Hume's devastating critique of metaphysics, Kant in his famous Critique of Pure Reason [22], concluded that there are elements or even rules of pure reason under which objects of the empirical world must be subsumed. If we see something that we think of as a triangle or a circle, there must be the very concept of triangularity or circularity within our mind. How else could we interpret anything as such by just perceiving or getting visual, empirical inputs. But what is considered to be a priori should then be independent of our experience. Indeed, there is analytic a priori as a concept that is already sort of built-in in our minds, and independent of any experience. Therefore, there is no analytic a posteriori, and although those truths can always be confirmed by our experience, they do not depend on it. According to Kant, we can recognize analytic a priori as logically necessary truths such as “All bachelors are unmarried“. But this will hardly tell us anything new about the world since “unmarried“ already is a part of the term “bachelors“. So the real knowledge about the world comes from synthetic judgements. Those judgements usually depend on experience and are built inductively, and we can go from case to case and then decide whether some generalization or a statement is true or not. Obviously, there is no reason to think of such statements as true or false without relying on empirical evidence. However, Kant thinks that synthetic can be both a priori and a posteriori with the former being a true philosophical problem. So where does he find synthetic a priori? For him it reveals itself in perceptual forms of time and space, and in categories such as quality, quantity, relation and modality. Consider this example: one is seeing an object. Let us suppose that object is a shovel that was already mentioned. What can we know about it? It could be black, brown, or yellow. It might be smaller or bigger, with a handle made of wood or metal. But each of those features can be known only in a synthetic a posteriori way because none if it is in any way already contained within the term “shovel“. However, one may be quite certain that the same shovel will fall if thrown from the roof of a building. Also, one can know for sure that the shortest way between its handle and its bottom part is a straight line. There are certain features that can be applied to any object and that makes them a priori but for Kant they are not part of analytic propositions. Shovels as such do not hold those features within themselves; they belong to broader laws of nature, mathematics or (Euclidean) geometry. And that is the true nature of Kant's synthetic a priori.

So, when McCulloch is concerned with how we can know universals or, differently put, how we can know we are perceiving a triangle, it actually means dealing with Kant's notion of synthetic a priori. We see a geometry form that has three sides and we know it is a triangle. We subsume empirical data under the general rule that, by itself, does not depend on that same empirical data. So if we talk in terms of pattern recognition what we do is we receive some visual inputs as empirical data and subsume it under some particular schema with it being a schema of circularity, triangularity etc. But what the described mechanism can really do is to give a binary answer, a 1 or a 0; yes, it is a triangle or no, it is not a triangle. The nature of any of the possible answers does not change even if we add a bit of fuzziness. Answers that scale from 0 to 1 are not binary but are missing the same thing as those that are; they do not differentiate the very concept of a triangle from its appearance. And what Kant is trying to prove is the existence of a transcendental schema which enables us to intuitively apply concepts to the appearance of an object. For Kant, that schema exists in pure thought, but for Pitts and McCulloch, according to How We Know Universals, “could exist in the brain, in specific neural circuitry“ [20]. Here we can notice a shift of understanding universal schemas and a priori from thought to neural systems. That shift was in a way confirmed in What the Frog's Eye Tells to Frog's Brain. By understanding perception in a more subtle way, McCulloch and his colleagues pointed out that it is dependent of the synthetic a priori:

By transforming the image from a space of simple discrete points to a congruent space where each equivalent point is described by the intersection of particular qualities in its neighborhood, we can then give the image in terms of distributions of combinations of those qualities. In short, every point is seen in definite contexts. The character of these contexts, genetically built in, is the physiological synthetic a priori. [7, p. 1958]

This synthetic a priori that is present in McCulloch's work Arbib brilliantly relates to animals' (including humans') behavior and biologically rooted entities that enables better development and adaptation to the world:

Moreover, we now understand that much, if not all, of spatial behavior of animals is controlled by their brain-body-environment interactions, and that if these rest on both 'nature' and 'nurture' then the nature is not of a priori structure, but rather a contingent structure shaped by evolution through natural selection. Such an 'innate, species-specific nature', moreover, is not directly expressed in adult behavior but, rather, sets a developmental pathway whose unfolding may be more or less influenced by the experience of the organism […] This leads to look at schemas not as immutable objects expressive of a priori principles but rather as biologically rooted entities which evolve and develop to better adapt the behavior of the animal, and the thought of the human, to its world. [20, p. 205]

So how does this relate to Fodor and Danto in terms of perception? The way I see it, McCulloch's work leads to refutation of some of the important claims about perception made in Fodor's understanding of perception and then, consequently, in Danto's claims. Those are: 1) we cannot think of brain/mind as modular, 2) top-down or bottom-up theories do not tell us everything we need to know about perception, we should think about it in a more subtle way 3) process of perception can avoid including central executive process but it is dependant on several brain regions and 4) perception is not informationally encapsulated and computationally autonomous – it distributes activity over a layered network and always includes at least something that Fodor might have thought of as cognitive or interpretative elements, schemas under which we subsume empirical data. But, as has been pointed out, these schemas do not have to be regarded as a priori in the way Kant understood them. It is possible to explain their existence through natural selection, evolution, and mutation to help organisms to better adapt to their environments. And that includes dealing not only with syntactic but also with symbolic/semantic propositions outside the central executive system which can be possible even for neural networks as McCulloch saw them [23]. In other words, semantics and symbolic representation is always present in each of our subsystems, including the visual subsystem. In this way we have to think of perception as being influenced by our knowledge and interpretation, even when it does not include central processing. In terms of AI, it means that vision has a history and that history depends on generalizations and propositions that are, at least in a way, semantical and can be learned.

Instead of conclusion: Where do we stand today?

Perception obviously does not work as Fodor tried to explain. Even if some form of modularity can be accepted, that would have to be structural, not functional modularity. That would account for “structural decompositions of the brain“ which “referred to physical structures as modules“ [15, p. 333]. Even McCulloch used this kind of modularity in the so-called RETIC model which tried to show how different modules interchange in giving executive orders; whichever module has the most important information is the commanding one [see 15 and 20]. This approach suspends hierarchy and keeps decision making, and the way I see it interpretation too, close to its primary source of input. In a way, what we might consider is that if by perceiving we manage to form new knowledge, it may be on disposition to the subsystem that received the input. Even so called low-level systems can make use of knowledge about the possible objects in the environment, something that would otherwise be accessible only to high-level systems. In turn, this might prove that subsystems which receive visual inputs can also use the perceptual schemas one is equipped with via natural selection and adaptation. Remember, with biologically rooted schemas which are mutable and which evolve through experience, organisms can adapt better – both the behavior of all animals and the thought of humans. Does this mean there is a schema for contemporary art objects? No, of course not. I lack both the courage and evidence for such a claim. But maybe it is possible to support a “weaker“ claim. In a way, the eye actually has a history and its history is directly dependent upon McCulloch's a priori in terms of genetic and evolutionary mechanisms. That being said, it is no surprise that Arbib [20] referred to developmental psychologist Jean Piaget's use of the term schema. His view of schemas followed the developmental process of human beings in which early instinctive sensorimotor schemas led to further development of schemas for language, logic, and abstract thought. Those schemas match exactly the genetically given a priori, depending on the actions an organism should be able to conduct, and bridge what we see as functions and what actually happens in the brain as a neural state. Taking those actions into consideration, there is no reason not to think that we simply learn. Our actions, by which we try to interact with objects and other human beings, might prove to be unsuccessful. In this case we haven't really recognized the situation and the object we were trying to recognize. What was once a schema for an everyday object can thus turn into a schema for an everyday object that might be a work of art, and that decision can be left to our subsystem. It is simply an improvement of our abstract thought and interpretations that can be found outside the central system. By no means am I even trying to say that the context in which an object appears is what makes it a work of art or that it is not so. It is possible that this is the case, and debates over primacy of contextual and formal properties of artworks are far from over. In a sense, where I can agree with naturalists in the so-called perception debate is that human beings have always seen the world in the same way. I think Danto is right when he claims so, that processes in our retina are still the same as they used to be before modern and contemporary art and ways of representing. However, I am certain that Fodor's modularity is an unsolid foundation on which he should not have built his claims. Finding a pigeon within us may be a completely wrong approach, and maybe he would have done better with frogs. Or differently put, Danto would not lose anything important if he had allowed interpretations to be made, at least to some extent, by our subsystems.

What I am pointing out is that the perception debate in philosophy of art hit a dead spot and became too oversimplified to be informative. There are ways to describe how vision works for both animals and humans. The real question is what is going on after visual inputs arrive. And that is a thing far more complicated than just splitting the brain into different modules that are considered to have different functions with no interaction whatsoever. Now, pigeons and frogs cannot comprehend works of traditional or contemporary art, they have no mechanism to do so, and the reason they lack it is because they are not supposed to comprehend it. So the mechanism described by Danto should not have been limited to contemporary artworks. But we can comprehend them. It might be the case that we were supposed to comprehend it, and our genetically given structure would eventually let us learn how to deal and interact with such objects. Unfortunately, in all of the books and papers I have read concerning perception and its role in philosophy of art, not once have I stumbled upon Warren McCulloch's name. The debate is still on and it keeps focusing on whether cognition is influencing our visual system or not, yet it is still limited to central processing and whether our visual subsystem is enclosed or not. There is no word on distribution of activity, layers of neurons, role of schemas, or anything we might even consider as a priori, no matter the exact form it might take. Perhaps the progress and development of AI will bring these connections out in the open. After all, if the eye has a history, at least in the described way, further development of neural networks and genetic algorithms might bring us not just closer to seeing and recognizing patterns more clearly, but also to the possibilities of recognizing contemporary artworks that formally do not differ from everyday objects. Principles of self-organization, activity distribution, and schemas by which such objects could be recognized might then prove to be far more interesting and far more important than they were in recent years, at least in philosophy of art. Personally, I am very fond of Danto's ideas. But there is a reason why some philosophers of art do not consider ready-mades or everyday objects as art. After all, there is more than one problem with hypothetical art and the indiscernible gallery of everyday and art objects the way Danto suggested. With their ontological status being the problem of semantics and symbolical propositions, he should have addressed those propositions and the way they work rather than relying upon Fodor's functional modularity which has proven to be inadequate. McCulloch's work was far more interesting and informative for the perception debate in philosophy of art than the one made by Fodor.


[1] Danto, A. C. 1981. The Transfiguration of the Commonplace: A Philosophy of Art. Harvard University Press. Cambridge MA.

[2] Danto, A. C. 1974. “The Transfiguration of the Commonplace“. The Journal of Aesthetics and Art Criticism 33(2): pp. 139-148.

[3] Beardsley, M. 1982. The Aesthetic Point of View. Cornell University Press. Ithaca NY.

[4] Zangwill, N. 1999. “Feasible Aesthetic Formalism“. Noûs 33(4): pp. 610-629.

[5] Adajian, T. 2016. “The Definition of Art“. The Stanford Encyclopedia of Philosophy. Zalta, E. (ed.). <https:

[6] Danto, A. C. 2001. “The Pigeon within Us All: A Reply to Three Critics“. The Journal of Aesthetics and Art Criticism 59(1): pp. 39-44.

[7] Lettvin, J. Y., Maturana, H. R., McCulloch, W. S., Pitts, W. 1959. “What the Frog's Eye Tells the Frog's Brain“. Proceedings of the IRE: pp. 1940-1951.

[8] Pitts, W., McCulloch, W. S. 1947. “How We Know Universals: The Perception of Auditory and Visual Forms“. The Bulletin of Mathematical Biophysics 9: pp. 127-147.

[9] Danto, A. C. 2001. “Seeing and Showing“. The Journal of Aesthetics and Art Criticism 59(1): pp. 1-9.

[10] Carroll, N. 2001. “Modernity and the Plasticity of Perception“. The Journal of Aesthetics and Art Criticism 59(1): pp. 11-17.

[11] Konečni, J. 2017. “Impenetrability of Visual Perception: Implications for Aesthetic Analysis“. Theoria 2 60: pp. 5-18.

[12] Nanay, B. 2016. Aesthetics as Philosophy of Perception. Oxford University Press. Oxford UK.

[13] Lotz, C. 2009. “The Historicity of the Eye. A Phenomenological Defense of the Culturalist Conception of Perception“. Phänomenologische Forschungen: pp. 79-94.

[14] Fodor, J. A. 1983. The Modularity of Mind. MIT Press. Cambridge MA.

[15] Arbib, M. A. 1991. “Modularity and Interaction of Brain Regions Underlying Visuomotor Coordination“ in: Garfield, J. L. (ed.) Modularity in Knowledge Representation and Natural-Language Understanding: pp. 333-363.

[16] Robbins, P. 2017. “Modularity of Mind“. The Stanford Encyclopedia of Philosophy. Zalta, E. (ed.).

[17] Churchland, P. M. 1988. “Perceptual Plasticity and Theoretical Neutrality: A Reply to Jerry Fodor“. Philosophy of Science 55(2): pp. 167-187.

[18] Anderson, J. A., Rosenfeld, E. (ed.) 2000. Talking Nets: An Oral History of Neural Networks. MIT Press. Cambridge MA.

[19] McCulloch, W. 1960. “What Is a Number, that a Man May Know It, and What Is a Man, that He May Know a Number?“ General Semantics Bulletin 26/27: pp. 7-18.

[20] Arbib, M. A. 2000. “Warren McCulloch's Search for the Logic of the Nervous System“. Perspectives in Biology and Medicine 43(2): pp. 193-216.

[21] Rohlf, M. 2020. “Immanuel Kant“. The Stanford Encyclopedia of Philosophy. Zalta, E. (ed.).

[22] Kant, I. 1999. Critique of Pure Reason. Cambridge University Press. Cambridge UK.

[23] Rescorla, M. 2020. “The Computational Theory of Mind“. The Stanford Encyclopedia of Philosophy. Zalta, E. (ed.).