1
A Revised Notion of Epistemic Consciousness
by Alexander Almér, Göteborg University
”Suppose by some cosmic accident a collection of molecules formerly in random motion were to coalesce to form your exact physical double. Though possibly that being would be and even would have to be in a state of consciousness like yours, that being would have no ideas, no beliefs, no intentions, no aspirations, no fears, and no hopes... This because the evolutionary history of the being would be wrong” (Ruth Garrett Millikan)[1]
The passage quoted above tells us about three relations. First, it says that meaning does not belong to the physical order but instead to the evolutionary order. Secondly, it says that consciousness is supervenient[2] on physical states. Your physical double "would have to be in a state of consciousness like yours". Thirdly, it tells us that there is no essential connection between meaning and consciousness.
Millikan works in a naturalistic framework. I propose that we explore the possibility of simultaneous application of at least two scientific frameworks, at the same time, in our examination of the relations between meaning, matter and consciousness. What I am after is a revised notion of epistemic consciousness that allows for Millikan’s physical double to have thoughts etc. I propose that we retain a naturalised notion of meaning, but supplement it with a non-naturalistic notion that fits a revised concept of consciousness. The (Cartesian) concept of conscious thought, the object of revision, roughly states:
(i) Thoughts, percepts and sensations are typical conscious states.
(ii) A conscious state is epistemically transparent. If you consciously think that Q then you know that you think that Q.
(iii) This knowledge is infallible.
(iv) This knowledge is necessarily private. That could mean several things. I here cite the epistemic interpretation. If something is private it allows, in principle, no intersubjective access. If something exists independently of there being a mind (a subject) relating to it in thought or otherwise, we shall say that it has objective existence. The concept of privacy formulated in terms of access does not entail non-objectivity, but it is hard to conceive of a world populated with objects, independent of my mind, that no one but I could (conceptual necessity) have access to.
Millikan designs a theory of intentionality that she claims is incompatible with any concept of consciousness containing any of the properties listed above. She substitutes a theory of concepts based on abilities to identify thoughts (tokens) and thereby contents of thoughts for the classical thesis that the meaning of our thoughts are known a priori, by reflection.
In the first section I describe Millikan's model. In section two I discuss different scientific frameworks and argue that some of them are compatible with one another. In the third section I discuss in what direction we should look for a notion of meaning fitting a revised notion of consciousness. Putnam[3] argues that there is an insight in pragmatism. I take up that thread. I end up saying a few things about which properties consciousness might have.
My aim is not to produce a theory but to point at a certain problem and set out a program for solving it.
Teleosemantics
The purposive order
Lets start with a brief account of Millikan's notion of proper function[4]. I will not go into details, but I shall give enough material to show how the notion of intentionality is fashioned out of the notion of proper function.
The basic intention is to create a tool for classification of functional devices. Millikan's route is essentially non-actualistic, i.e. functional categories are not defined according to what the things in these categories actually do. Membership of a functional category thus doesn’t depend on actual dispositions or causal roles of the device.
The main reason for Millikan not to accept actualistic accounts of functions is that they cannot offer a decent notion of malfunction. According to Millikan such a notion presupposes a distinction between having a function and performing a function. A malfunctioning device has a certain function, the very same that it fails to perform. Ergo if actual dispositions or actual roles would define the function then by definition a device couldn't have a function without being able to perform it.
What counts for something being a function, or should we say, for a device or a state to have a function, is what it historically has been selected for or designed to do. Biological structures "designed" by selective pressure are paradigmatic cases. That a device has a proper function means, roughly, that its ancestors[5],
(members of the same functional category), have had some effects which have accounted for —in accordance with cateris paribus laws— their proliferation and in the end for the existence of the current device. This doesn't mean that these effects are actually produced or even that the device is capable of producing them. Two important concepts are Normal explanation and Normal condition. Normal conditions are those conditions that account for, according to a Normal explanation, the historical cases where a proper function was performed. The Normal explanation covers historical cases of actual performances and shows how these are covered by natural laws given certain external conditions and certain properties of the devices.
Millikan speaks, perhaps metaphorically, of the evolution taking part in a purposive order[6], a special part of the causal order, distinct from the physical ditto. I shall adopt that expression. Purpose, in this context, means proper function. The purposive order is in the natural world. Its laws are ceteris paribus laws, derived from universal natural laws by addtion of a set of actual conditions which define the circumstances under which the law is operative[7]. According to this picture intentionality of thought is an aspect of our biological functional system .
Intentionality as part of the purposive order
Millikan defines a basic form of representation termed "intentional icon". There are two sorts of intentional icons, indicative and imperative. They can be inner representations or sentences in a natural language. They are also exemplified by other kinds of signs, such as bee-dances, animal or human warning signals etc. The term "representation" is reserved for a specific kind of indicative intentional icon. The human belief system is a representational system. We come back to that later.
An intentional icon has a function in relation to two other functional items. These are a producing device and a consuming device. The consumer activity is modified by its contact with icons. Icons in turn are produced to modify consumer activity in systematic ways . Consumer activity either results in the performance of its proper function or it doesn't. The same holds for the producer of the icon and for the icon itself. The icon, the producer and the consumer are developed as a system of interrelated functional devices. None of them works without mutual co-operation.
What makes an intentional icon entitled to the characterisation intentional, is the relation between the icon and the Normal conditions for performance of the consumer's proper function. An indicative icon is supposed to map —in accordance with a specific projection rule— the Normal conditions for proper consumer activity. "Supposed to", in this context, means that if the icon is to be able to perform its proper function it has to map according to the rule. The consumer activity can do its job properly only if the icon maps. This is not, however, to say that the icon's purpose is to map. Its proper function is to modify the consumer and the mapping rule tells (us) which the Normal conditions are.
The mapping rule that connects certain properties of the icon with certain "supposed" conditions of the world is the sense[8] of the icon. The conditions mapped (if it works properly) are called "real value". This means that a true sentence corresponds according to its sense (mapping rule), i.e. it has a real value. This basically equals Millikan's correspondence theory of truth. A true sentence maps a world affair and a referential term maps a constituent part of that affair. Referential terms have sense as well as sentences, but they work properly only in a sentential context. There is an essential difference between sentences and referential terms. The sense of a sentence does not presuppose that there is something to which the sentence is supposed to map. All sentences are not true. Referential terms, on the other hand, must have a history of being parts of true sentences where they did in fact refer, That is what gives them sense.
The act of identifying
The consumer of the icon need not "represent" the sense of the icon to "interpret" it, neither does any other part of the system. The internal mechanism of the interpreter has been selected for responding in a specific way relative to certain properties that the icon can have. This mechanism has a function and we can speak of internal "rules" for consumption, but according to this view they are not semantic rules. They tell us (or the system) nothing about the mapping rules. This, however, doesn't imply that Millikan claims that we cannot know what we think. She rejects the notion that we have a priori infallible knowledge of the meaning of our thoughts. She also denies that meaning is something internal to the system. Instead she suggest a weak form of knowledge of one's own thoughts. As mentioned above, Millikan reserves the term "representation" for a certain form of intentional icons. One of the things that sifts out representations is that they can perform some of their functions only if they are identified qua meaning. The system performs an act of identifying[9] whenever it bases its production of new icons or action on more than one intentional icon that overlap in content. A paradigm case is the coordination of activity based on perceptual representations in different sensory modes. To manipulate an object that we e.g. both see and feel the system has to know that the tactile and visual representations overlap in real value.
Another example of acts of identifying is mediate inference, where, in the premises the same representation is instantiated twice. The system —using different tokens of representations with the same real value to perform an inference— performs an act of identification.
Suppose that you on two different occasions think of a rose-bush in your garden. Late in spring you look at the bush and think "This sort is a late flowerer; it will not bloom until June". May comes and you see the first flowers and think "It blooms in May; I was wrong about that bush being a late flowerer". To correctly infer that you were wrong you have to identify the "rose-bush" parts of your thoughts.
The ability to identify a certain representation is termed "having concept". There are several ways of identification included in one ability. We can recognise a certain person by sight or we can identify him by description. If we see him it could be his way of walking, his clothes, his face or whatever makes us able to identify him. Also, nothing prevents us from having two concepts of the same entity or short-lived concepts as of your glass at a specific party.
Millikan's solution to the identification problem is not to posit a second order thought, directed at temporally or otherwise distinct thoughts (but not necesarilly distinct as to contents). Neither does it have anything to do with a mediating factor[10] as sense or mode of presentation. Rather it is a designed ability of the system to handle different representations as having the same real value. Among the normal conditions for handling them that way is sameness of real value. Millikan wants us to think of concepts as know hows. There is no guarantee of successful performance for any ability. There is nothing inside showing me (or anyone else) if I actually think. First, a person who performs what she believes is an act of referring, may be wrong; namely if there is no external object Normally related to the act. From this follows that she does not know a priori if she thinks. Secondly, we cannot know a priori if two of our thoughts or utterances have the same content. Our abilities to identify work only under Normal conditions. Thirdly, we cannot know a priori if we mix two contents in same thought. Of course, this is the desired result for Millikan.
Scientific frameworks
Intuitions
I would say that there are two strong intuitions that many of us share, that are captured by Millikan's concept of intentionality. One tells us that the truth value of what we believe or claim is determined by how things in the world in fact are. What is true is not relative to theories, societies, minds or methods of verification. Truth seen as correspondence captures this intuition better than any other known notion of truth.
The other intuition states that we should try to understand all sorts of things without bringing in notions that fall outside the field of, or are incompatible with, the laws of natural science. This intuition, in turn, is captured by Millikan's program of naturalising intentionality.
Here we are presented with yet another intuition which seems totally incompatible with Millikan's teleosemantics. It brings us back to the issue of epistemic consciousness. Take the rose-bush case again. Your attention is caught by a wonderful rose, its colour, all the nuances, the ingenious organisation of petals etc. The properties, the rose itself, everything is right there before your mind. Isn't that an example of conscious knowledge of meaning? It is granted that we know that some of our thoughts are not so clear (the same goes for our saying) but that is not the issue. Millikan's notion of concept gives no inkling of how to explain the "feeling" of having the rose before the mind. It just says: "It cannot be there".
Why naturalise?
I intend to dwell upon naturalisation and other scientific frameworks for a while. Take intentionality: what motivates a naturalistic approach? Well, think of the basic types of situations which comprise semantical aspects: communication, perception, action and thinking. All contain a physical course of events. The communicator moves his tongue, a sound signal is produced which changes the physical (or physiological if you wish) state of diverse nervous systems. We can depict the physical courses in a similar way for the other meaning-relevant situation types. In addition to the physical structure of these situations there is a meaning structure. This structure can also be depicted as a course of events, with a possible exception for perception where one may say that the only meaning relevant event involved is the occurrence of the percept. Communication it sets off with meaning "in" the speaker and winds up with meaning "in" the hearer. Or it starts with reasons, beliefs and purposes and ends up with actions (meaningful behavior). This description does, however, not tell us how meaning comes into the world. If meaning is not a part of the physical world, how come that we can communicate, perceive, think and act only in connection with physical courses of events? Is the connection just a guise? Now ask the same questions about consciousness: add the intuition (those of you having it) that perception and thought sometimes are conscious. We are obliged to ask how consciousness enters the world.
There are basically two different approaches to these issues. One is to deny the connection between physical and semantical aspects of the situation types. One could claim that meaning is fundamentally separated from the physical. Perhaps one posits an immaterial consciousness capable of producing meaning. A defender of this position must account for the apparent connection, between the meaning structure and the physical structure, in the situation types listed above. If she doesn’t we shall not bother listening to what she says. The other way to handle the question is to try to explain the connection. There are at least four basic ways to do this.
Physicalism
First, one may try to physicalise meaning, which basically means that one defines everything that has to do with meaning in physical terms or in terms ultimately reducible to physics. The same goes for consciousness. This is the strongest form of naturalism.
Multiple-order theory
Secondly, one can handle the issue in terms of different orders of reality. The physical order is a structure consisting of the flow of energy in space and time. The physical order is the basic order, because all matter is structured this way (at least we tend to suppose so). This structure is the one described by modern physics. The intentional order could be another structure of energetic flow in time and space. Not all matter is intentionally structured. This provides a way to understand Millikan's idea that intentionality is part of the teleofunctional or "purposive order" found in evolution. Of course, those parts of space-time slices containing evolution are physically ordered as well. We could perhaps say that the purposive order is supervenient on the physical order. Again, the same goes for consciousness. One can thus try to account for consciousness in terms of a supervenient order. Millikan considers her strategy to be naturalistic. I propose we extend ”naturalism” to cover all multiple-order theories.