Introduction to Biosemantics Josef Behr August 16, 2011 Abstract This paper deals with Ruth G. Millikan’s naturalistic theory of meaning, explaining such concepts as intentionality and language by biological means and in the end trying to reduce the intentional description of the world to a biological one. The intention though is not to provide a complete overview over the theory and everything related to it but rather to give a short and understandable introduction into the framework and the notions of natural, intentional and linguistic signs, on which grounds many ideas of hers should be easier to comprehend. 1 Prologue Ruth G. Millikan, born 1933, is a professor emeritus of the philosophy of mind, language and biology at the University of Connecticut. Since her 1984 book “Language, Thought and other biological categories” she has been defending, refining and expanding her theory of meaning, blurring the lines between philosophy, psychology and language, and integrating such highly debated subjects as intentionality or the origin of meaning in human language and thought. In 2002 the book “Varieties of Meaning” was published, which is a result of the Jean Nicod Lectures she gave in Paris, 2001. Great thing about this book is the concise and clearly structured explanation of the naturalistic framework Millikan developed. On the other hand, it is hard to read sometimes, because many concepts and new terms are introduced and set into relation to many others such that following the thoughts behind the written words is not easy. The goal of this essay is to provide an overview over the base concepts and ideas of Millikan’s theory for the interested reader and those already familiar with her writings as well. As an original theory depends highly on its terminology and concise definition as well as explanation of it, I will start with a structural classification of this theory and provide some background knowledge in the first part and then explain the terms at root of Millikan’s evolutionary framework, namely and most importantly those of “Natural Signs” and “Purposes”. Thereafter I’ll go on and describe the integration of the notion of intentionality into the proposed theory and following that a short introduction on how language is related to all of this. Especially regarding the part about intentional signs, my text is certainly shorter and hopefully a bit more accessible, since Millikan approaches the task from three different angles, which I tried to prevent and provide only the view Millikan herself employs and argues for. 1 2 A Teleosemantic Framework At the very first, one should be clear on some of the premises regarding the matter in question in this essay. Ruth G. Millikan’s theory is to be understood not as conceptual analysis, but as a theoretic model to explain meaning in its various incarnations. Meaning of words, of sentences, the meaning of a thought, a tool that is meant to be used and also the sound of thunder are different forms of meaning. As Millikan states in the Preface of Varieties of Meaning, she doesn’t think there is anything particular which is “common among the various things that are said to mean things”1 . What can, however, be explored, are the relations between those understandings of meaning. The meaning of something relates to either what its purpose is or what it signifies, Millikan states2 . The tool’s purpose is what it is meant to be used for, the sound of thunder is a signal of thunder and a word’s or sentence’s purpose is to signify the underlying thought transmitted by it. We will get a better look at the terms purpose and sign as well as their meaning in the light of Millikan’s theory later on. But first, I will try to draw the frame in which to put the theory. Franz Brentano argued that the property of thoughts and other mental phenomena to be of or about something, or to be directed at something, separates them distinctly from physical phenomena. Like I can have believes about the computer on my lap or about my flatmates. One important distinction characterizing intentional phenomena seems to be what he called intentional inexistence. That is, a thought or belief (or, for that matter, anything else which is said to be intentional) can be about something which doesn’t exist. About some shiny, fast, new computer, I’d rather have to write on for example. Of course, I could be thinking about a specific computer, but my thought isn’t about the computer being anywhere else but here in front of me at this exact moment. That situation most certainly does not exist.3 The notions of intentionality and especially of intentional inexistence pose a great problem to philosophers trying to naturalize the mental realm, whereby Naturalism4 denotes the assumption of a real, physical world (ontological realism) paired with the idea that everything concerning the mind can be explained by the natural sciences (only methodological naturalists claim that this is the one and only way to learn about it5 ). A Naturalist theory of meaning has to explain how it is that things mean something, how mental states can be intentional and how they can be about something which does not exist at all, if there is to be a natural relation between representation (sign) and represented (signified). One of these questions, namely the last one, is answered by Teleosemantic theories. Teleosemantics are functional (telos, gr.: purpose) theories of (mental) content (semantics: the linguistic study of what words and sentences mean) that have in common the understanding of a representation as something which’s function it is to represent. The function of the underlying system that produces representations is to bring representations to life, independent from the actual physical realisation of this system. If this system does not function properly (in the sense of it failing to do what it was designed to do), it produces representations which don’t represent anything at all. The same goes, of course, for intentional representations. False in1 see [Mil06], ix 2 see [Mil06], ix 3 see [Mil06], pp. 63-65 4 after all, Millikan was a student of Sellar’s and calls herself his “daughter”, see [Mil04] 5 see [Pap09] 2 tentional representations, ie. the imagination of a unicorn before the inner eye, are then not really representations at all. Their content doesn’t possess any natural connection to the world which would be hard to explain if what is represented didn’t exist. That’s the clue of this functional point of view regarding intentional representations (and related phenomena). It gives an explanation of why signs or representations that ought to be about something, can be representing something that’s not there. Like, as Millikan underlines this point, a can opener is still a can opener, even if not functioning properly6 . A representation can in the same sense also still be a representation (because it was its proper function to be one), although it doesn’t succeed in representing anything at all. The reason, Millikan argues, that this idea of a representation representing nothing but still being a representation, because the producing system was supposed to produce representations, sounds paradox, is due to the missing linguistic distinction between trying to represent and succeeding to represent. The purpose (which is claimed by teleological theories to be the constituting factor) in both cases is to represent something, but sometimes this purpose is just failed to be fulfilled and nothing is represented at all. At the base of such theory need now come the explanation of why it benefits an organism to produce intentional representations and what exactly constitutes an intentional representation. That it benefits an organism to produce intentional representations is a requirement laid upon the constructive part of the teleosemantic theory by its biological assumptions. If producing (intentional) representations wouldn’t be helpful in any way, there’s no reason for them to exist in the first place, if the idea is that evolutionary processes eliminate traits not helping or even hindering an organism. If the system bringing about intentional representations was selected for 7 , it has some effect that benefits the organism. This is, in Millikan’s terminology, the purpose it was selected for.8 Also the relation between the non-intentional representations, which Millikan calls basic representations, and the intentional ones has to be looked at. This will be the scope of the following sections of this essay. 3 On Purposes In the normal usage of many terms, we make a difference between their literal and metaphorical understanding (and, as we will see later on, Millikan’s theory can account for such differences in meaning). For instance, something which is used to be sat on, may be called a chair, although, if not used for sitting on it, one may not be inclined to call it that, but rather say it’s a box, a piece of wood or whatever it is you actually sat on. Then this object may be viewed as being a chair, but only - as we say - metaphorically. But, what is a chair, literally, then? Even with physical objects the line between those two categories is blurry. If mental objects like concepts are involved, it gets even more fuzzy. Could you, on the spot, say what the essential difference between a metaphorical and a literal idea may be? The same, as Millikan argues, is true for the notion of purpose. Although a purpose is normally understood as an explicit human intention ("It is my purpose to go to the lecture tomorrow" or "It is the intended purpose of this essay to explain a specific theory"), which is the literal understanding, there are many cases that blur 6 see [Mil06], p 9 7 In contrast to being just “selected” for some effect correlating for no reasons with the traits existence 8 see [Mil06], p 68 3 the line to a more metaphorical meaning of purpose. The most obvious case may be posed by biological reflexes. If I have the explicit intention to hold open my eyes (for example, the doctor is attempting to put eye drops in to stop a burning sensation), one can easily see that it is my purpose as a person to hold the eyes open (as long as I’m keen with getting eye drops in). But there is also a biologically developed reflex of my body to close the eyes as soon as some external object gets near enough to possibly damage them. The doctor and his instruments certainly are an external factor with the ability to hurt my organism, so there is also a purpose on a more subpersonal level which crosses with that on the personal one9 . The question raised hereby is if either one of those purposes is more real than the other one. Maybe the conscious, full-personal purpose is? In Millikan’s understanding of the concept of a purpose, there is no differentiation between real and metaphorical understanding. She argues that "no interesting theoretical line can be drawn between these two kinds of purposes"10 . Instead, she sees the world as layered whereby on any layer from the biological description of things to the mentioned full-personal view of a human purposes can be found. The higher levels of purposes are nontheless no special kind of purpose but rather “made up out of intertwined purposes”11 at the lower levels down to mere biological purposes. A good example Millikan gives to illustrate this idea is the taste for sweet foods. On a biological level this taste has been selected for by evolutionary processes to reinforce the eating of sweet things, which provide lots of calories to the organism. The behavior resulting from this reinforcement which is selected on a genetic level, has a purpose which results from the genetic purpose of reinforcing sweets-serving behavior, namely that of eating sweet things containing calories. The reason for this is probably the strong correlation between sweet foods and high calories in the history of animals on this planet. But do we eat sweet things with the goal in mind to get calories which we need? Certainly (most often) this is not the case. So there is yet another layer of purposes, for example the learned and conceptualized purpose of sweets serving my explicit conscious desire to eat sweet foods, at which level it’s not relevant if there are calories in there or not. This point gets interesting though, as soon as the purpose of sweets silencing my desire for them crosses with the explicit purpose of me not wanting to get fat or ill from too much sugar (which in our modern western culture is the pre-dominant source for a sweet taste)12 . According to Millikan, the prominently discussed weakness of will is nothing more than explicit purposes being un-fulfilled, because they cross other purposes which are for whatever reason fulfilled instead. Millikan seems to think - and I’d agree with her - that the notion of a biological purpose is clearly understandable in an evolutionary context. But having the greater task of naturalizing the mental realm in mind, the crucial question regarding purposes is how the biological purposes relate to the seemingly non-biological, mental ones. These explicit desires, intentions and so forth13 , these “mentally represented purposes”, Millikan states, "represent the conditions of their own fulfillment"14 . Their purpose is to produce exactly that what they represent. Hence, a 9 see [Mil06], p 3 10 see [Mil06], p 3 11 see [Mil06], p 3 12 As was pointed out to me this “explains the usage of calorie-free sweeteners” very well, at least considering the weight aspect 13 Intentional phenomena as introduced above 14 see [Mil06], p 8 4 connection between the notion of purpose and that of an intentional mental state is established. A desire for eating chocolate cake has the purpose of producing a state of the world in which a chocolate cake is eaten by the one having this desire. The idea behind this is that by means of evolutionary mechanisms and natural selection an apparatus or mechanism was chosen for its capacity to bring about those explicit representations, which sometimes lead to their own fulfillment and hence further some underlying biological goal. This cognitive mechanism was selected on a genetic level, but the actual mental representation was certainly not. It should also be noted that of course no requirement of actual fulfillment of every desire or intention is needed. It seems, Millikan says, that the "capacity to develop and to act on desires [...] [has] been selected for only because desires are sometimes fulfilled and, of course, sometimes do represent means to fulfillment of our biological interest"15 . Although the selection of consciously represented purposes happens on another level than that of purely biological ones, namely by experimenting in thought and in turn forming those explicit desires, goals and intentions, but nontheless the ability to do this is a product of natural selection and biological purposes undeniably play a large role in forming those explicit intentions. Returning to the chocolate cake, it seems obvious that this desire is somehow rooted in the biological purpose of the good taste of calorie-rich food, even if the person having this desire is not aware of this fact. Or, on yet another level, the chocolate cake may unconsciously remind her of the one her mother made on special occasions when she was a kid. Millikan doesn’t stop at the explicit human level of purposes, but goes even further and proposes that concepts, beliefs, artifacts and even language are results of biological processes and serve a purpose for which they were selected on a cultural level. These memes - cultural genes, if you like - like specific rituals in any culture or even language were again not one by one selected for because they benefit an organism, but rather the capacity to imitate, replicate and modify existing ideas was selected, because - remember the requirements of a Teleosemantic theory - the resulting memes more often would do so than cross with the genetic purposes. An example Millikan gives is the arrow, which benefited the producing organism by providing a better chance at killing an animal to eat or in a fight. The purpose of this meme then in turn is also the purpose for which the meme gets replicated. Whereas in those cases for most memes the purpose is at core still a biological one, serving the purposes of lower levels as well, Millikan distinguishes those memes having special purposes not found on lower levels, namely cooperative purposes. These are special because they seem to have been produced by a mechanism facilitating social coordination, helping two or more individuals at the same time by allowing them to coordinate their individual behaviors16 . The most important kind of cooperative purposeful memes are conventional elements of language. The purpose they possess makes up their meaning which they signify at the same time. Conventional meaning of language signs is a meaning which serves its purpose if speaker and hearer alike "are trained to respond with and to it in ways that have some stability, each given the expected performance of the other"17 . Me uttering the words "Help me!" serves no cultural purpose unless the one hearing those words knows how to act or react accordingly to them. Millikan also gives some thoughts about the phenomenon of conformity in hu15 see [Mil06], p 9 16 see [Mil06], p 19 17 see [Mil06], p 26 5 man cultural behaviors by stating her suspicion of a genetic disposition to conformity for the sake of social coordination18 . In contrast to others she does not believe that conformity is a means of spy detection as to find out who acts differently from the other members of a specific cultural society. She argues that most social interaction is not competitive but rather cooperative and benefiting everyone19 . She also mentions that in her opinion no theory of the mind of other members of a species is needed for an individual to predict the other individuals behavior. For this, she argues, the past observed behavior is foundation enough to predict, for example, if someone will be on time. If this observed person was on time in most of the past cases, we most likely assume that she will be on time the next time, too. If she wasn’t, we’ll expect her not to be, whatever the actual reasons may be. Any assumption on the observer’s part about these reasons will do, no matter if it is correct or not. 4 Natural Signs The second important notion at the core of Millikan’s Biosemantics is that of a natural sign. Fred Dretske, developing an information theory of meaning, claimed that a sign or signal carries natural information if and only if it is the case that due to logical necessity and natural laws the sign’s occurence allows to grasp information about the sign’s source with a probability of exactly one20 . As Millikan exemplifies this would mean that the smell of a fox is a sign for a rabbit, carrying natural information about there being a fox nearby and hence danger, if the occurence of this specific smell is by natural laws bound to the occurence of a fox and the possibility of being eaten by it. But, as she also points out, no natural law forbids the advent of a new species smelling like foxes but at the same time being harmless to rabbits.21 Another problem for such a theory employing logical necessities is the representation of individuals. No natural law connects the current chancellor of Germany and Angela Merkel, resulting in there being no possibility for a sign carrying information about an individual. Consider again the smell of an individual rabbit. Nomically it is possible that another rabbit or even a completely different animal smells like this individual and therefore according to Dretske’s theory no natural information about an individual could be carried by such sign. There’s yet another problem as Dretske seems to require specific channel conditions to hold for a sign to carry specific natural information, but does not elaborate on how an organism would acquire knowledge about those conditions since they are not part of the sign itself. To avoid those problems and at the same time provide a good theory of natural information that explains how organisms can acquire knowledge about some part of the world (signified) by means of another part (sign), Millikan proposes a notion of natural signs that relies on statistical considerations instead of strict natural laws. She distinguishes the new notion from Dretske’s one by calling his kind of natural information context-free and the one she employs local information, because it considers both “conditions at the source of information and on channel conditions”22 . Also, for animals to be able to actually learn about this information, there must be a nonaccidental connection between the information carrying sign and its source, 18 see [Mil06], p 20 19 see [Mil06], p 21 20 see [Mil06], p 32 21 see [Mil06], p 33 22 see [Mil06], p 34 6 at least in a specific domain trackable by the animal. In most cases this connection is probably a causal one. Black clouds are a sign of rain because they are the cause of that rain. But there’s no requirement given for the connection to be of any specific kind other than being not there by accident. In the case of bacteria finding oxygen-less waters (which they need to survive) by means of magnetic orientation away from the magnetic north, no causal connection between the magnetic field and lesser oxygen in deeper waters exist. Instead, in the local domain swimming away from geomagnetic north is a way to get in deeper waters, because due to natural laws the north pole stays in roughly the same relation to deeper waters over time and deeper waters correspong to lesser oxygen.23 A natural sign in its simplest thinkable form is then just something from which one can learn something about something else, because there is in fact a real connection between the two which can be traced in thought. As Millikan correctly states, this is “at root an epistemic notion”24 , for the rabbit having learned that in its living environment the smell of a fox means it has to run or hide if it wants to avoid being eaten. But this simple definition of a natural sign does not account for the learning. Only with prior knowledge (the smell correlates with the presence of a fox, foxes eat rabbits, etc) about the relation between sign and signified can the information be useful. Hence, an important requirement for the notion of a natural sign in the light of evolution and historical learning is to include a way of explaining why its occurence can be used by organisms to successfully learn this connection. Quite obviously a single occurence of such a natural sign would not suffice to explain how by natural selection or learning mechanisms an organism can use this sign to acquire knowledge. An isolated group of rabbits in a region where there were never any foxes probably wouln’t know to run from the smell of a fox, since its occurence has no relevance as it couldn’t have learned or be genetically primed to react in a specific way. Therefore the relation between the sign and its signified has to be recurrent. The statistical correlation has to persist in its reference class, which Millikan calls the natural domain, in which this relation and its recurrence holds. For example, meeting a heavily-bearded man in my flat (being one possible natural domain I keep track of) is a sign for my flatmate being at home, but this inference is no longer true as soon as I leave the flat. Out there a man with a beard no longer signifies my flatmate, since the natural domain has changed. A rabbit may never leave its habitatual natural domain and never know that somewhere there may be foxes that would like to eat it (imagine pet rabbits), hence in its domain the relation between the smell of a fox and danger is not given. With these considerations at hand, a more precise and useful definition of a natural sign can be given, which accounts for them and additionally shows some interesting properties, as we’ll see. First, the already mentioned recurrence of the relation between sign and its signified in a specific domain of possible signs and their signifieds is important for such a natural sign to bear any relevance for an organism, since it must be able to learn from or be evolutionary adapted to it. Second, saying that for example black clouds mean rain is certainly kind of weak regarding its expressiveness. When and where will the rain begin to fall? How much rain will there be? Just having something which relates to something else (the natural connection) and can be used as a sign of that second thing is not very specific. It also leaves out the domain in which sign and signified occur. Therefore natural signs can 23 see [Mil06], pp. 44-45 24 see [Mil06], p 37 7 be better thought of, as Millikan calls it, “structured world affairs and the things of which they are signs are also structured world affairs”25 . Returning to the rabbit and the fox once again, the smell of a fox at a specific time t and at a place p may be a sign of there being a fox at this specific time and place. As can be seen, there exists a correlation between variables of the structured world affair which is the sign and those of the world affair which is the signified. The time of the smell corresponds to the time of the fox’s being there, the location from which the smell arrives corresponds to the location of the fox26 . In conclusion, a natural sign consists not only of the recurrence of sign and signified, but in the recurrence of the specific relation between them, which constitutes a semantic mapping of variables of the former to those of the latter, similar to a mathematical function mapping input variables to output variables.27 Therefore, natural signs can be understood or defined as semantic mapping functions from structured world affairs to other structured world affairs whereby this relation recurs locally within a specific domain trackable by an organism. In the case that such natural sign has more significant variables than time and place, Millikan finds it more natural to call it a natural representation instead. In the first case a natural sign may be thought of as a picture of the affair it signifies, showing time by time and place by place. In the case of what she now calls a natural representation though, if more significant variables are involved, the representation may be considered to be carrying (local) information about the affair it represents.28 What kind of information this is or, in other words, what sign/representation means is determined according to its significant variables and their structure. Simply put, a lot of heavy-looking black clouds in Osnabrueck on Sunday mean rain later on the same day or the next one or no rain in Osnabrueck, but in a town nearby29 . Hence, the representation’s structure determines its meaning in the way that changes in variables of the represented affair imply changes in variables of the representation. If the clouds are white instead of black or appear not on Sunday but on Tuesday, their meaning changes. 5 Properties of Natural Signs Due to this structural determination of a natural representation’s meaning and a sign’s being an affair of the world representing another affair of the world yields two interesting properties which are normally attributed to conventional linguistic signs30 , but not natural signs. For the sake of explaining intentionality and language in biological terms this is an important factor, since conventional linguistic signs 25 see [Mil06], p 47 26 These variables of signs may be called reflexive, if they represent themselves (size stands for size and place for place, etc.) 27 see [Mil06], p 49 28 Yielding in effect a hybrid understanding of the theory as a picture theory of meaning in the case of simple natural signs and an information theory in the case of natural represenations, see also [Mil06], p 50 29 This may seem strange at first, but the black clouds at time t are definitely a sign of either of those possible outcomes. Which one it actually was a sign of will be seen as soon as it rains or if one tracks these specific clouds long enough. 30 Conventional linguistic signs are the signs used in natural language by convention as in the normal meaning of a word or utterance, the one it was culturally selected for to mean. The open question of how exactly natural signs are related to conventional linguistic signs as well as to the nature of linguistic signs in general will hopefully be answered in the following sections. 8 are a special case of intentional signs and being able to characterize natural signs by these properties at this stage of theory construction can only be useful in doing the step up to intentionality. One of these properties is productivity in the sense that due to changing or recombining existing signs, a new meaning of this new sign can be derived. This is thanks to the structural architecture of the signs and representations proposed here. Although the meaning of a sign is given by its structure as explicated above, it is not derived from possibly independent meanings of its parts or aspects (previously called variables) put together, but the meaning of the parts is derived from the complete meaning of the complete sign in the context of its specific sign system and domain. Changing one or more of those aspects of a given sign then yields a new meaning, hence productivity.31 Putting the black clouds above Munich instead of Osnabrueck changes their meaning from being a sign of nearing rain in Osnabrueck to a sign of nearing rain in Munich. Regarding conventional linguistic signs like words and sentences this productive property is often called compositionality. The meaning of a linguistic sign is taken to be derived from the meaning of its parts and a change to the parts changes the meaning of the whole. Exchanging a word in a sentence may completely alter its purpose. Where “Give me a hand” is a call for physical help, “Give me a hint”, although only one word changed, describes something completely different. Compositionality is in this sense just a special case of productivity in signs due to the way they are structurally build. The other important property of the natural signs as Millikan defines them is embedding. Embedding describes the ability of one sign to be included in another sign. It’s not quite obvious as to how our locally recurrent natural signs exhibit this property. Acknowledging though the understanding of a natural sign as a structured affair of the world, it seems easily conceivable that a sign A of a sign B, which in turn is a sign of an affair C, is at the same time itself a sign of that more distal affair 32 C. One might say that there is a route from affair A to affair C, given by the application of the the semantic mapping function from A to B to the one from B to C. This is what Dretske called the Xerox Principle33 . Viewed this way, the route between two signs can be arbitrarily long. For example, consider the case of the fox. In principle it is simply put if one says that the smell of the fox is a sign of danger for the rabbit. In reality there are particles in the air entering the rabbit’s nostrils coming from a nearby fox. These particles may be signs of different parts of a fox and thereby signify the existence of those parts at this time and place. The parts in turn are in their composition a sign of the fox’s existence, which is a sign of danger for rabbits. This more complex route of signs is sure expandable. It shows though the fact that an organism needs not to know or even realize an intermediate sign on a route, as long as the semantic mapping preserves the respective meaning. It also exemplifies where the embedding can be found. Omitting the time and place variables - since time and place of the smell corresponds to time and place of the fox and therefore time and place of the danger to the rabbit - the smell is a sign of a part of another sign (the fox being the sign for danger). It is of course not needed that every sign on the route is recognized to interpret a sign. A new-born rabbit sensing a nearby fox may already be genetically predisposed to react in a certain way corresponding to behavior expressed in the face of danger, although it never before sensed a fox. Here 31 see [Mil06], p 48 32 see [Mil06], p 54 33 Probably - I’m guessing here - after the name of the company that developed the first plain photocopier, see [Wik11]. In contrast to Millikan Dretske required for the principle to hold that a hundred percent of the information had to be carried over from A to C, see also [?] 9 an important aspect becomes visible. Encountering a sign often requires additional knowledge and tracking of the sign’s domain to be able to correctly interpret it. The image of a combination of shape, size, structure and other properties that hits my retina can be a sign of my flatmate if they were caused by him, but to know that it is my flatmate, this specific individual person, I must track the domain in which this specific combination is actually a sign of him. If I’m in Hong Kong and know that my flatmate is at home, the image may represent his Doppelgänger. This retinal representation of an individual, as Millikan puts it, is “perfectly ordinary”34 . As it was already the case with the property of productivity, embedding is also a feature of natural language and hence conventional linguistic signs. Here this property is shown by what Millikan calls defining descriptions35 . Defining descriptions are expressions like “my flatmates”, describing who I’m talking about in a defining manner as only specifically defined persons match this description. Other examples include “the president of the university” or “more intelligent than Sally”, all of which have in common this feature of describing properties of a person or thing without mentioning who it actually is who I live with, who currently is the president of the university or how intelligent Sally or the person compared to her exactly is. Nontheless there is someone who is my flatmate or the university’s president and by using the defining description one can refer to this person and at the same time signify other properties of the person which are associated with the defining description alone. It seems that in the sign that is the defining description another sign, namely that of the actual person referred to, is embedded. Neither one, the actual person or the defining description is in that case a complete sign of the other. When using the description “president of the university” one can still talk about that person, even without knowing either what a president or a university is, if it is known who is referred to by it, or who the person is who is the president.36 6 Intentional Signs Now is a good time to take a step back and have a look at the bigger picture at hand. In the beginning, the question was what the constituting factors of meaning are. The idea was that meaning in principle is linked to the notions of purpose, since things mean something whenever they have a purpose, and sign, as one can say that a sign of something else means whatever it signifies.37 By clarifying the understanding of a purpose as being not only “real” in the case of an explicit human intention, but accepting the world as being purposeful on many levels, most importantly on a biological level, keeping evolutionary processes in mind, we can talk about purposes of sign-producing systems. Also, a link between intentional phenomena and this notion of purpose as intentional states having a purpose and representing their own fulfillment conditions38 is established. Remember then the teleosemantic framework in which Millikan’s biosemantic theory is to be integrated through constructively explaining where meaning and intentionality come from in the first place. Then the main idea behind the previous construction of the notions of natural sign 34 see [Mil06], p 55 35 see [Mil06], p 59 36 see [Mil06], pp. 59-60 37 see section 2 and [Mil06], pp. ix-x 38 Millikan explicitly states that she wants to avoid using the term “truthmakers” as to not imply specific signs of subject-prediate structure only, see [Mil06], p 47 10 and natural representation as locally recurring signs with their properties of productivity and embedding is, that organisms benefit from acquiring, learning or adapting to the carried local information as a means of survival. It is the purpose of the system adapting to a sign to have a certain effect on the organism. Learning that the smell of a fox means danger is essential for rabbits living in the respective habitats, because it leads them to run and hide. This sign serves the purpose of helping them survive. On the bottom line, a biological system was selected for evolutionary, which allows those rabbits to learn and interpret these signs - or rather, what they mean, for what they stand. For the system’s being selected for, only the condition of bringing in the costs its development generated is laid out. It benefits the rabbit if it can detect a nearby fox rather than it would benefit the rabbit having no means of fox-detection at all. The fox-detection system should, though, function properly and therefore detect a fox whenever there actually is one, otherwise there would be no reason for it to survive.39 One has to dinstinguish though the purpose of a trait from the mechanisms by which it fulfills its purpose, causing some effect that in turn somehow helps the organism’s survival. The predominant mechanisms used by a trait are called normal mechanisms in the light of Millikan’s theory. Taking for example the eye blink reflex, the purpose of some biological apparatus selected for by evolutionary means may be to cause the eye lid to shut in preventing foreign objects to penetrate the eye. The normal mechanism by which this is accomplished is the one where the system causes the eye to close because there actually was a foreign object, since the system was selected for doing exactly this. It is also possible, Millikan argues, “that my eye-blink, though only caused by a passing shadow, might nontheless succeed accidentally in keeping a piece of sand out of my eye.”40 This then is not the mechanism by which this happens normally, but the sytem’s purpose is nontheless fulfilled. But in the case that the eye closing is caused by the grain of sand and not a shadow, it functions properly and the system’s causing the eye lid to close is a natural sign of a foreign object nearly entering the eye. The idea then is, regarding intentionality, that organisms make intentional signs out of the local information carried by natural signs as explained before, and this local natural information41 is what gives meaning to intentional signs or representations42 . Put another way, if a system that produces intentional representations does so by its normal mechanism, “then the intentional representations are basic representations - whatever ’basic’ representations are taken to be.”43 And of course, in this biosemantic theory, basic representations (the non-intentional ones) are taken to be natural representations. It follows from this, that it is not a proper function of any system to actually produce natural representations, but this is rather the normal way by which intentional representations become true, if they are natural signs at the same time. We still need to answer the question what exactly constitutes an intentional sign and what the relation to natural signs is. Millikan introduces her understanding of intentional signs by once again distinguishing two historical aspects. On the one hand there are systems producing natural signs as a side effect of some other func39 see [Mil06], p 72 40 see [Mil06], p 69 41 Again in contrast to Dretske’s context-free natural information, which Millikan argues to be nearly non-existent and much less user friendly, see [Mil06], p 71 42 From now on these terms will be used completely interchangeably, since, as Millikan puts it, “to represent, in the base sense, will then be to be a natural sign”, see [Mil06], p 67 43 see [Mil06], p 69 11 tion they fulfill, for example the calluses showing where a wear has been. It is not, however, this production of a natural sign which the system producing the calluses was selected for, but its ability to protect the skin. On the other hand, there are systems producing natural signs and their purpose is actually to do so. Like, as Millikan exemplifies, a hen clucking when she found food and as a result having her chicks follow her to the food source. The important difference between these cases is that in the first one the production of a natural sign was not intended, whereas in the case of the mother hen it was “purposefully produced in order to serve the chicks as a sign of food.”44 This sign, produced because it serves a purpose which in turn benefits the organism producing it (the hen) as well as the organism or system interpreting it (the chicks), is then an intentional sign. From this introduction to the understanding of intentional signs it becomes clear that an intentional sign is a cooperative sign. Meaning that it was produced by one system for use by another co-evolved or designed system, leading to a benefit for both systems. Considering the eye-blink reflex once again, in the case of it working by its normal mechanism, the signal causing the eye-lid to close is an intentional sign and at the same time a natural sign, carrying local information about a foreign object trying to enter the eye, but in the case of it fulfilling its function by accident, the signal is just a true intentional sign as it was purposefully produced to indicate the presence of an object, not a natural sign, as there was no object. What kind of sign a representation is depends solely on its history.45 Let’s go into more detail on intentional signs. As defined here, an intentional sign is a sign produced by a producing system, which may be an organism or part of an organism, with the purpose of being understood by a consuming system, which then in accordance with the meaning of the sign does something proliferating the producer (otherwise there would be no reason for the producing system to produce this kind of sign) and the consumer alike. Both the consumer and the producer system are part of the normal mechanism by which the sign-producing system works and a sign “will be true or satisfied only if it maps onto some world affair in accordance with a definite mapping function determined by a history of joint successes of producer and consumer (or their ancestors).”46 In the case of the hen and her chicken, we can imagine that over time the specific clucking sound a hen makes when finding food co-evolved with the interpretation of that clucking as a sign for food by the chicken. Maybe the clucking was just a natural sign of food in the beginning, outer expression of internal processes of the hen and then chicken began to understand it as a sign of food, leading as a result to the hen to actively seek expression of this fact to help her children survive, which in turn proliferates the hen as her genes have a higher probability of being passed on. Where most theories of meaning and mental content consider only one type of intentional representations as important, namely descriptive intentional representations. These are intentional representations that correspond to world affairs and, if they were produced by the producing system’s normal mechanism, are hence natural signs. They change in accordance to change in the world and the purpose of such an intentional sign is to guide the consuming system or organism by performing some task proliferating both systems in the end. But in addition to descriptive intentional signs, Millikan considers two other kinds as equally important. The sec44 see [Mil06], p 73 45 see [Mil06], p 73 46 see [Mil06], p 79 12 ond one are directive intentional signs that correspond to the world in another way. Their purpose is not to guide a consumer by representing and changing with affairs of the world, but to represent a state of the world which the consuming part is responsible to bring about. If the consumer succeeds in producing an affair of the world corresponding to the directive intentional sign, it should again benefit both organisms that it did so. In the case of inner representations, the directive representations have the purpose to guide the construction of an affair of the world corresponding to it and this is the reason an organism has these directive representations as inner signs. The third and new kind of intentional representations Millikan calls pushmipullyu representations47 . These are especially interesting, since they seem to form a more primitive kind of intentional representation than the purely descriptive or directive ones. It is in fact both at the same time, descriptive as in the hen’s call telling where and when food is found, and directive as in the same call being only satisfied if the chicken succeed in following the sign and bringing about the affair of the world of them approaching the food. These pushmi-pullyu representations seem to be kind of a basic intentional representations from which in further development the descriptive and directive ones as special cases were derived. As Millikan argues, this kind of intentional representation is also the most prominent one among animal signals. “For example, bee dances tell at once both where the nectar is and where the watching bees are to go. The vervet monkey’s leopard, snake, and flying predator calls tell what kind of predator is near and direct the response appropriate to that predator.”48 To end the considerations regarding intentional signs here, the relation or better, the distinction, between natural signs and intentional signs should be examined more thoroughly. It was said that an intentional representation is taken to be a basic representation which in this biosemantic theory is then taken to be a natural representation, whenever the system producing the intentional representation did so by its normal mechanism. But it should be noted that the natural sign and the intentional sign are still not one and the same. In principle the hen’s call for example is a natural sign of many things at the same time, but it is an intentional sign only with respect to the affair(s) it was selected for to represent. In this example those are the current affair of food being at the place of the hen (descriptive aspect) and the future affair of the chicken moving towards this place and eating the food (directive aspect). Or in a more abstract example, the triangle drawn on a piece of paper is a natural sign of someone having drawn this specific structure on a piece of paper and more, but intentionally it may be a sign of the concept of triangularity alone. There is no problem with that. Only the information that proliferates both the producer and the consumer is per definition the intentional representation, whereas the natural sign is also a sign of any affair en route to other, more distal ones, it stands for. The intentional sign is that part of the natural sign for which it was selected to be used by the consuming system or organism. Another example in natural language that Millikan gives is the sentence “It is raining” whose “memetic function, derived compositionally from the combined memetic functions of its significant components, is to produce beliefs that it is raining, not beliefs that speakers believe that it is raining”49 . 47 According to a footnote on page 77 in [Mil06] this term comes from a “charming double facing creature by that name” by Hugh Lofting 48 see [Mil06], p 81 49 see [Mil06], p 83 13 7 Linguistic Signs Having a good understanding of intentional signs and their base in natural signs, Millikan continues by examining linguistic signs as a special case of intentional signs. As could already be seen with natural signs is that they show properties found in linguistic signs, namely productivity, in the case of language given by what is called compositionality 50 , and embedding of one sign in another. So what are linguistic signs, can they be natural signs? Linguistic signs are cooperative intentional signs that memetically evolved in group of conspecies and proliferated by imitation and reproduction because they serve communicative purposes. The idea again is that, through time and evolutionary processes as well as learning, conventional language signs gradually emerged from natural signs.51 This emergence can be imagined as a beginning at preparatory movements of animals in reaction to certain circumstances, natural signs of these circumstances (for example the rabbit’s thumping in case of danger). Sometimes it may have been useful for a cooperating organism (another rabbit) to know what a sign meant and at the same time was benefitiary for the sign-producing organism to intentionally produce such sign (the rabbit’s thump helps the other rabbit’s, because they know of the danger, and benefits the thumping rabbit itself, too, since the other rabbits running away may get the predator confused for long enough to escape). Thinking this further, it is conceivable that grunting and other phonemes uttered by humanoids as a byproduct of other actions became useful in “mutually beneficial social activities such as cooperative hunting, warfare, the creation of environmental structures [...] and so forth.”52 The natural signs gradually became cultural memes, proliferated by imitation and reproduction for the reason of serving a cooperative purpose beneficial to the speaker as well as the hearer. These conventional purposes of language forms correspond to their public meaning in a language community. The linguistic signs persist because the response of hearers to them encourages speakers to reuse them, adapting to an environment in which this sign serves a specific purpose. Hearers are reinforced to respond in a certain, conventional way to the linguistic sign as this behaviour aids them. For example, responding to the utterance “Careful, there’s a hungry lion behind you” in a conventional way aids the hearer, since he won’t be eaten by the lion (probably). And for the speaker it is encouraging to use this sign for its conventional purpose as long as hearers respond to it accordingly. Of course, to fulfill its memetic purpose normally 53 , the public language sign has to map onto world affairs. Either it maps to an existing world affair in the correct way and it is a descriptive intentional sign (“There is a man behind the window”), which in turn, as it fulfills its purpose by its normal mechanism, is also a natural sign, whose meaning depends on the domain in which it recurrs being used for this purpose. It can also map onto future world affairs if it is a directive intentional sign (“Don’t do that!”) and has as a purpose to bring about a state of the world corresponding to it, or it can be an intentional sign of both, a pushmi-pullyu sign (“I hereby pronounce you man and wife”).54 There is also another way as to how words and sentences can acquire purposes 50 Compositionality as a special case of productivity yielded by structural architecture determining meaning in general, see [Mil06], p 51 see [Mil06], pp. 104-105 52 see [Mil06], p 104 53 As in “by their normal mechanism which they were selected for” 54 see [Mil06], pp. 105-107 14 apart from the public meaning of a linguistic sign. This purpose is given by the speaker’s actual purpose he or she had in mind when producing the linguistic sign. This derived purpose, as Millikan states, can conincide with the conventional public function of a linguistic token or it may cross with the conventional public meaning, depending on how the speaker uses it. If it is used conventionally, everything works as explained above, the public language sign’s own purpose for which it was replicated is the same as a speaker’s purpose in using this specific sign. But the nonconventional usage of linguistic tokens seems to pose a problem at first, since the requirement for intentional signs was that producer and consumer alike were coevolved or designed to function together using the sign. The question then is if and how speakers and hearers of linguistic signs are designed to use signs in a nonconventional way, where previous encounters of said token don’t necessarily help in understanding. A speaker using a word in a nonconventional way has to influence the hearer in a way which is in accord with what the speaker intended. There has to be an inbuild capacity to understand this derived purpose, too, in addition to the conventional one given by historical usage. Otherwise the sign wouldn’t be an intentional sign as proposed in this theory and as a result wouldn’t carry any information at all. Millikan argues then that the system seeking communication in a nonconventional way and the system spontaneously able to understand such nonconventional use have co-evolved in speaker and hearer, designed to cooperate for communication purposes, whereby this ability seems to be largely species-dependent. The nonconventionally used linguistic sign then is an intentional sign “carrying content intented by the speaker”55 . If these nonconventional language forms get to be used more often and proliferate for their serving a cooperative purpose, they become recurrent intentional signs which is then a form of a local conventional sign.56 8 Epilogue It’s obvious that although one might try to clarify terms, notions and concepts, the natural overlap and gradual transition from natural sign to intentional sign and nonconventional linguistic sign to a conventionally used one don’t always allow a clear and sharp distinction. Nontheless I hope the main ideas at the base of Millikan’s biosemantic theory of meaning were explained in an understandable manner. There is a lot more to the theory and I guess man questions remain (yet) unanswered. Regarding the notion of intentionality, Millikan has given a biological explanation of this phenomenon, which already accounts for some characteristic properties often ascribed to intentional mental states57 . For instance, they have a content, namely the conditions of their fulfillment. Millikan wouldn’t say though, that it is a “propositional content” as she explicitly distinguishes the memetic purpose and conditions of satisfaction/fulfillment of an intentional sign from its semantic mapping function and states that the notion of a proposition “hovers [somewhere] between”58 them. They are also, at least if true or satisfied, systematic (in their compositionality) and productive, since they are then also natural signs and inherit their properties. In the 7th chapter on Intensionality the phenomenon of differing intension of two intentional signs while having the same extension is examined and explained by the 55 see [Mil06], p 108 56 see [Mil06], p. 108 57 see e.g. [Bec08], p. 300 58 see [Mil06], p 92 15 difference between satisfaction conditions and semantic mapping function of the signs.59 In a bigger context, the theory seems to provide a neat reduction of what Dennett called the intentional stance60 to a functional view having its roots in biological purposes and meaning acquired from nature. I can only recommend the interested reader to dive into the original writings either via her homepage61 , where drafts of three books and many independent papers can be found which answer a lot of open questions, or via the published books themselves. “Varieties of Meaning” itself holds many more chapters on perception through language, the linguistic debate over semantics vs. pragmatics and other linguistic phenomena, as well as elaborate considerations regarding inner intentional signs, representations and (non-)human thought, which would unfortunately have had this essay’s scope exploded. References [Bec08] Ansgar Beckermann, Analytische einführung in die philosophie des geistes, Walter de Gruyter, 2008. [Den87] D. Dennett, The intentional stance, MIT Press, 1987. [Mil04] R. G. Millikan, The father, the son and the daughter: Sellars, brandom and millikan, http://www.philosophy.uconn.edu/department/millikan/father.pdf, 2004. , Varieties of meaning - the 2002 jean nicod lectures, The MIT Press, [Mil06] 2006. [Pap09] David Papineau, Naturalism, http://plato.stanford.edu/archives/spr2009/entries/naturalism/, 2009. [Wik11] Wikipedia, Xerox, http://en.wikipedia.org/w/index.php?title=Xerox&oldid=443445296, August 2011. 59 see [Mil06], pp. 87-100 60 see [Den87] 61 http://www.philosophy.uconn.edu/department/millikan/ 16