Hilary Putnam’s causal theory of mental content determination has long been influential in the literature. I will show here that, in addition to the various arguments against the theory on the grounds that the argument is unsound, it rests on some problematic assumptions and devices. The Twin Earth thought experiment is contrived in order to generate the desired intuition. The underlying assumption about the nature of the grounding relation between representing vehicle and represented object is plainly false. Causation is not the only available option for a grounding relation.
These will be discussed in three sections. First I will introduce the concept of representation in general as well as mental states and their role in mental representation. Then I will consider Putnam’s theory and its problems. Finally I will present a structural theory of mental content determination in order to show that there is a viable alternative to causation.
Mental States and Representation
In order to fully understand what is at stake in this argument it is important to understand the nature of the battlefield. As such there are some peripheral matters to describe and define. These are mental states, representation, and mental representation.
Being largely outside of the scope of this essay, the nature of mental states is assumed here to be purely physical. That is to say that for a subject’s mind to be in any given state is simply for that subject’s brain to be in a given state. What that state is most likely to be is a pattern of neuron firing rates across a neural network. Whether that neural network is instantiated within a human or other animal brain, a computer, or some other substrate is unimportant here.
Representation in General
In simple terms, representation is a ‘standing in’ of one thing for another. However it will be important to note that this stand in must have the function of standing in, there can be no accidental standing in within a representational system, and it must perform this function within some general scheme of things standing in for others. (Clark and Toribio 1994) Calculus provides strong examples of this. Consider the equation F=MA. The standard Newtonian equation of force. In this equation each letter stands in for an observable value of the environment; F for Force, M for Mass, and A for Acceleration. They stand in as such because that is their function. They are in the equation for the purpose of representing those values. The scheme of representation here is the calculus itself, it is a branch of mathematics that takes values from the environment and transforms those values into others which may not be quite so readily observable.
On the above definitions then mental representation becomes the standing in of patterns of neuron firing rates across a neural network for objects, states of affairs, relationships, etc., where those patterns have the function of standing in and are doing so within a general scheme of standing in. From modern neuroscience we know that the brain can be conceptualised as many of these networks working together, extracting information from various sensory modalities, passing the information back and forth, and using that information to navigate in an environment.
The following discussion will centre on how these states come to carry that information, and what makes a particular brain state be about something else.
Putnam wants to say that, if two people can be in the same mental state in the narrow sense but be referring to different objects then some part of what determines the intension of the mental state must be external. (Putnam 1975) To unpack this idea we will need some tools.
A mental state in the narrow sense can be taken to mean any mental state that it is logically possible for a lone mind to be in. (McGlone 2010) That is a mental state that does not assume the existence of any other mental states.
Intension is the term used interchangeably with the term ‘aboutness’ in order to describe the fact that mental states have some meaning, they are directed at something. This something towards which these mental states are directed is the represented object, also known as the intensional object. So when you think about water, water is the intension of that particular mental state. Those things which have intension, those mental states to which I refer, are the representing vehicles which will become very important later on.
So a thought about water is a particular mental state that does not assume the existence of any other mental states, and is directed at, or takes as its intensional object the substance the subject calls ‘water’. What Putnam wants to say about this is that you and I can be in the same mental state in the narrow sense with respect to the substance each of us refers to as ‘water’ and yet those mental states can have different intension or aboutness. If successful this theory would show that something external to the subject determines, at least in part, what its thoughts are directed at. What we will need to examine is the extent to which the theory is successful, to support this claim Putnam develops a set of thought experiments, the most important of which I will now examine.
A Journey to Twin Earth
Imagine that there is another world in some distant part of the universe. This world is exactly alike our own down to every molecule, Putnam says there is even a doppleganger of each of us on this planet. There is however one very important difference, in every instance wherever there is H2O here on earth there is instead some other substance with all of the same properties on Twin Earth. This substance instead has the molecular structure XYZ. XYZ falls from the sky as rain, fills the oceans and rivers, flows from the tap, and presumably makes up 70% of the Twin Earth human body. On this planet the subject to which I will refer also speaks English, they even call XYZ water as we do with H2O.
Now consider yourself and your doppleganger, when you think about water so does your counterpart. The intensional object of your thought is H20 and theirs is XYZ. A complication arises when instead of thinking about yourself right now you think about Oscar and OscarTE (the Oscar of Twin Earth) who live on their respective planets in the year 1750. At that point in time no one on either planet had yet discovered the molecular composition of water. Oscar has somehow managed to travel to Twin Earth. He encounters this liquid XYZ, he hears the locals refer to it as water, and he assumes it is water as he knows it. Putnam now wants to say that since the actual substance being referred to on Twin Earth is of a different molecular structure, when Oscar and OscarTE are having thoughts about water in 1750 they are having thoughts about different intensional objects the same as you and your doppleganger.
The strength of a causal theory is its ability to account for concrete particulars. C. S. Pearce whose work Von Eckardt (1993) builds upon said that the weakness of resemblance relations is that they are only capable of representing general classes of objects, they can represent apples but they cannot represent my apple. The entire Twin Earth experiment is designed to show that there is a difference in the extension of the word ‘water’ (and thus a difference in the object of thoughts about ‘water’) determined by the actual stuff in the environment of the speaker (or thinker).
There are various problems with this. I will consider only a few of them in the next section.
Problems with Twin Earth
This thought experiment is one of the class that Dennett (1980) would probably refer to as an ‘intuition pump’. This is a general type of thought experiment that I would describe as contrived in order to generate the particular intuition its author desires the reader to experience. In this case Putnam wants to exploit this far-fetched but logically possible scenario in order to make the reader think that there is obviously some external consideration in determining exactly what each subject’s water thoughts are about.
Given that it wasn’t possible in 1750 to determine the chemical composition of anything, it does not follow that ‘water’ meant H2O. To speak of chemical compositions in the context of 1750 mental states is nonsense. (Christensen 2001) Historically many liquids have been referred to as water (tears, urine, sweat, saliva, etc.) it also does not follow that water necessarily meant ‘the colourless, odourless, and tasteless liquid that falls in rain and fills the oceans and rivers’ in 1750. Even today there are various chemical compositions of water, D2O, T2O, HDO, HTO, DTO are all types of water. Within a typical sample of water that we might find on earth today there are usually several different types of water molecule as well as sodium chloride and other minerals. So it does not even follow that water now refers only to H2O. (Zemach 1976)
Before we turn to an alternative however we should first take a look at how Putnam found himself in the position where causation seemed like the only viable grounding relation between these objects and mental states.
It is generally accepted that there are only three candidates for a grounding relation in representation; resemblance, convention, and causation. Convention plainly does not have the power to ground meaning in mental states since a convention requires mental states for its maintenance, to rely on it would lead to an infinite regress of meaning being explained in terms of another state.
Left with only two options, Goodman (1969) made the claim that resemblance cannot be a viable candidate. This claim was because resemblance is symmetrical while representation is not. An object resembles itself completely and yet does not represent itself. A painting resembles its subject to the same degree the subject resembles the painting, yet only the painting represents its subject. Resemblance is ubiquitous, many things resemble others and yet they cannot be said to represent them.
Very shortly we will see that Goodman, far from raising a good objection to resemblance theories, has only pointed out an issue with dyadic theories of representation. We can now see that there are some significant problems for Putnam’s theory.
Structural Theory of Mental Content
O’Brien and Opie (2004) refer to their resemblance theory instead as a structural theory. This becomes an obvious move when considered in terms of second order resemblance as they describe it. This section is not intended as either a complete exposition of their structural theory, nor as a defence, but as an alternative to show that causation is not the only option as Goodman and Putnam would have us believe. In order to see how this undermines causal theories we will need to overcome Goodman’s objection, we do this by appealing to the triadic nature of representation. We need also to appeal to an abstract form of resemblance that does not refer to physical features since the brain, according to our current understanding of it, cannot resemble the property of redness for example.
The traditional conception on which Putnam based his theory was one of a dyadic relation between object and vehicle. It is because of this that Goodman was able to raise the objections already mentioned. But in the face of the triadic theory and its addition of the interpretant to the picture we are about to see that Goodman in fact has no grounds for objection.
For a representation to obtain there must be some object to be represented, this can be a thing, an emotion, a person, etc. There must be a vehicle by which the object is represented, this vehicle must bear some relation to the object, this vehicle can be a word, a picture, or in the case of mental representation a mental state. And the lynch pin of the theory is that there must be an interpretation, that is to say that there must be some subject for which the vehicle does represent the object such that the subject is brought into an appropriate relation with the object. (Von Eckardt 1993)
This overcomes the objection Goodman gives to resemblance as a grounding relation because it is no longer mere resemblance doing the work. It is the fact that the resemblance the vehicle bears to the object brings the subject into an appropriate relation to that object. The fact that A resembles B and B resembles A is irrelevant unless one or both of those resemblance relations brings some subject into an appropriate relation to the object. Once we have seen that not all resemblances are representations we can do away with the importance of ubiquity and we can see that, as described, the symmetry of the relation is also unimportant to representation. (O’Brien 2016)
Applying this to mental representation is no simple task though. As we are about to see it requires a degree of abstraction that takes resemblance from a sharing of properties to a structural relation.
Second Order Resemblance
Until now I have discussed first order resemblance. Those are resemblances of a property sharing nature. An apple bears a first order resemblance to a tomato in that they share the property of being red. But there are other resemblances which do not share such simple properties.
The simplest example of second order resemblance can be found in a mercury thermometer. The physical thermometer shares no first order resemblance with the ambient temperature. But it does bear a structural relation. The volume of the mercury increases as the temperature increases such that there is a one to one relation between specific volumes and temperatures. This mapping of values in one set to the values in another is where we find our structural relationship.
In the case of mental representation this would be a mapping of mental states to states of affairs in the world where the state of affairs is the object, the mental state is the vehicle, and the subject is the cognitive agent containing that mental state. Second order resemblance is the grounding relation between vehicle and object, but what of the relation toward the object that the cognitive subject must be brought into to complete the circle of the triadic theory of representation? This relation can be simply described as the subject behaving appropriately toward the object. So in the example of the thermometer, the ambient temperature is the object, the thermometer itself is the vehicle, Oscar has come along and looked at the thermometer, based on the information he gains he determines that he needs to wear a jacket on his adventure to twin earth. He is now behaving appropriately toward the object that is represented by the thermometer. (O’Brien and Opie 2004)
Before making any conclusions, let us first see where these new tools have left us with regard to the plausibility of Putnam’s theory, and causation as the grounding relation between object and vehicle.
Strengths and Weaknesses
We can clearly see from the thermometer example that at the very least certain kinds of particular can be represented through second order resemblance relations. What of two apples though? What I would ask of a causal theorist is ‘what does it matter?’
Our triadic story tells us that what is important is that the subject ultimately behaves appropriately toward the object. If there are two apples and I am unable to determine from resemblance which one is mine, then there is no need to determine which on can be labelled ‘my apple’. This returns us to the final objection I raised against the Twin Earth thought experiment. Since H2O and XYZ fulfil precisely the same functional role and are superficially identical, there is no need to determine which is which and as such there is no advantage to any requisite difference in mental states. Thus the primary advantage of causal theories becomes its primary weakness in that it now requires a large amount of explanatory baggage in order to explain how it pulls off this entirely unadvantageous differentiation of concrete particulars.
By now we have seen that the primary support for Putnam’s theory is built on a shaky foundation of intuitive contrivance, misinformed assumption, and a dedication to a dyadic approach. Once we point out that the thought experiment is an intuitive contrivance purely designed to extract the intuition that is needed for support of the theory and that this intuition is not necessarily a natural one the thought experiment loses all of its force. Add to this a refutation of the misinformed assumption that causation is the only option. Then, through the triadic approach to representation which emphasises the behaviour of the subject towards objects in its environment, we are left with the conclusion that Putnam’s “meaning of ‘meaning’” really has no… Meaning. Whether a structural theory of mental content determination is true or not, the very plausibility of it suggests that causation is not the only option. If there are other options and those options provide us powerful and naturalistic explanations then surely we must reject the bulky intuition based attempts of Putnam and all of his followers to claim that “Meanings just ain’t in the head!” (Putnam 1975 pp. 144)
Christensen, C. B. (2001) “Escape from Twin Earth: Putnam’s ‘Logic’
of Natural Kind Terms”, International Journal of Philosophical Studies, 9:2, 123-150
Clark, A. and Toribio, J. (1994). “Doing without representing”, Synthese, 101, 401-431
Dennett, D. In Reply to Searle, J. R. Minds, Brains and Programs. The Behavioural and Brain Sciences 3: 417-457, pp. 428-430
Goodman, N. (1969). Languages of Art, UK: Oxford University Press
McGlone, M. (2010). “Putnam on what isn’t in the head”, Philosophical Studies: An International Journal for Philosophy in the AnalyticTradition, 151, 199-205
O’Brien, G. (2016) “How does mind matter? Solving the content causation problem”. In T. Metzinger & J.M. Windt (eds.) Open MIND. (MIT Press) Vol.2, pp.1137-50
O’Brien, G. and Opie, J. (2004) “Notes Toward a Structuralist Theory of Mental Representation”, in H. Clapin, P. Staines & P. Slezak (eds) Representation in Mind: New Approaches to Mental Representation (Elsevier, Oxford), pp. 1-20
Putnam, H. (1975). “The Meaning of ‘Meaning’”, Minnesota Studies in the Philosophy of Science, 7, 131-193
Von Eckardt, B. (1993), What is Cognitive Science? MIT Press
Zemach, E. M. Putnam’s Theory on the Reference of Substance Terms, The Journal of Philosophy, Vol. 73, pp. 116-127