Notes on Kripke "Naming and Necessity," Lecture 3, Part 2 of 2 (pp. 136-155)

1. Continuing on p. 136, where Kripke is in the middle of discussing his distinction between a priori and necessary knowledge, he portrays the “simple" way we referentially fix the natural phenomena we directly observe with our senses by presenting it as an equation:  “Heat = that which is sensed by sensation S.” But isn’t heat the sensation itself?

Other natural phenomena he says are “identified as the causes of certain concrete experimental effects.” Okay - but then, such phenomena are "baptised" via description.

It would seem that the only objects that can be baptised without description are the immediate contents of perception, and here Kripke seems blocked by a failure to distinguish between the sensation and what it is a sensation of. This issue will come up at the end of the lecture, when he attempts to argue against identity theory.

Now we reach the third main point of Kripke’s argument: Once we have identified a natural kind (by virtue of paradigmatic cases, however that is supposed to be established), we can add additional members to the set by determining if they have properties which are “roughly characteristic” of the original cases, and he says these properties “need not hold a priori of the kind”(137). He says it is always an empirical matter whether any particular properties are necessary or sufficient for membership.

It follows that no description of any natural kind is a priori necessary. It follows that the meaning of a natural kind term cannot be given by a description. For if it could be given, then the description would be analytically true, and therefore a priori necessary.

Kripke wanted to undermine Frege’s distinction between sense and reference. (Russell took Fregean senses to be descriptions, but Frege took them to be modes of presentation. Kripke does not comment on this distinction.) For Kripke, a name has a meaning and a fixed referent. The meaning is supposed to be available by description, but we have just seen that this is not the case. So the meaning of names, and the relationship between meaning and description, is entirely unclear. Names refer according to properties that transcend our epistemic situation, and they have meanings which transcend any description of them. This makes names entirely mysterious, and does not account for anything.

2. Kripke’s fourth point is that “scientific investigation generally discovers characteristics [of natural kinds] which are far better than the original set”(138). This is odd, because the original set was not characteristics; it was objects. Science discovers the “essence" shared by the objects, the precise structural nature of a thing. Kripke’s reference to “structural traits” here is reminiscent of the distinction between form and matter. His claim is that science discovers a pure form, so that accidental (material) properties can be distinguished from what is necessary.

He gives the example of “heat is molecular motion” here. He says that science “seems” to use a “type of property identity” that is associated with what he calls “necessity,” but not a prioricity or analyticity. In other words, he is claiming that when science attributes properties to a kind, it does so with the force of metaphysical necessity. But is he trying to distinguish this from semantic necessity, because that would make it a matter of analyticity. So he does not want to say that “heat" means “molecular motion.” He wants to say that molecular motion is a metaphysically necessary property of heat.

Of course, scientists don’t talk this way, and it does not seem to reveal a truth about the way the actually do talk. Scientists today define heat as thermal energy, or more specifically as the thermal energy transferred between systems. It is not simply “the motion of molecules." We can say,

(H_def) Heat is thermal energy transferred between systems.

This is simply a definition of the word “heat.” Thermal energy is somewhat loosely defined as the kinetic energy of atoms in a substance—that is, the energy produced by the motion of atoms within a substance. From this we can deduce that heat is produced by the motion of atoms. This is knowable from the definitions of “heat" and “thermal energy.” It is therefore an analytic truth.

Kripke says no, this is not an analytic truth. We need science to tell us that heat is thermal energy transferred between systems, because it is an empirical fact, not an a priori truth. So Kripke sees the scientific definition of “heat" as establishing a metaphysical connection between a kind (the thermal energy transferred between systems) and an object of experience fixed by the referring term "heat." Thus, he takes H_def to mean that "heat" refers to “the thermal energy transferred between systems” in all possible worlds, even though we need empirical evidence to support the claim.

But why can’t we say that “heat" means “the thermal energy transferred between systems”? It certainly seems to mean that, as per H_def. Simple substitution also seems to work:

(H1) Can you feel the heat coming from the engine?

(H2) Can you feel the thermal energy transferred between systems coming from the engine?

If somebody kept using “thermal energy transferred between systems” in a scientific discussion, somebody could rightly complain, “Why don’t you just say ‘heat’? It means the same thing!” 

So the metaphysical necessity Kripke is after seems to be analyticity. I suppose Kripke wants to avoid difficult questions, such as, “Did ‘heat' refer to heat before the science of thermodynamics was developed”?

Such questions are only difficult if our answer matters for some particular purpose. We can assume that most of what people referred to with the word “heat" was the same as what people refer to as heat; except that now our understanding of the processes has deepened. The meaning of the word “heat" has changed, but the referent has largely stayed the same. As a Fregean would put it, thermodynamics allows us to regard the same referent under a different mode of presentation. The scientific definition expressed by H_def is not the common definition of “heat,” but they both (usually, I presume) refer to the same phenomenon. We can say that H_def expresses a different concept of heat. It has a different Fregean sense than our everyday understanding of heat.

The difference in Kripke’s approach is evident in the example of “the concept of fishhood”(138). In the past, whales were considered fish. Today scientists classify them as mammals. According to Kripke, scientists have not changed the meaning of “fishhood.” They have just revealed a flaw in older ways of thinking. The essence of fish and mammal have been discovered, such that “whales are mammals” is a necessary truth.

But isn’t it obvious that the definition of “fish" has changed? Did people who once believed whales were fish simply lack knowledge of whale anatomy and physiology? Or were they applying different concepts? The answer cannot be decided without investigation into how the term was once used.

3. Kripke’s fifth point: the discovery of new items can “augment" the “original sample” that fixed the referent of a natural kind term (138-9). Scientists apparently use the “rough characteristics” to find new items, and then decide that these new items can change how we think about the original sample. It follows that our determination of natural kinds depends on descriptions according to rough assessments of characteristics.

Kripke here comments on “the traditional disputes over primary and secondary qualities,” but only to claim that his theory offers an essential way of understanding them: first, he refers to his marks on property-identity, and second, his claim that our intimate phenomenological connection to certain objects (such as light) tricks us into thinking that the objects cannot be known apart from those phenomenological processes. Light is still light, whether or not anyone has any eyes. 

He has a lengthy footnote on p. 140 where he tries to explain this with the example of yellow. He says the referent “yellow" is fixed by description, “that (manifest) property of objects which causes them, under normal circumstances, to be seen as yellow (i.e., to be sensed by certain visual impressions.” So we do not fix the referent through our direct experience of yellow, but through a description? This is . . . curious. Also, Kripke says here that this referent-fixing description is a definition of the word “yellow" according to a standard of synonymy, and not according to a scientific discovery. But this proposed synonym of “yellow" includes the term “yellow,” so that can’t be right. Honestly, I am at a loss.

After this, Kripke introduces the notion of necessary a posteriori truth in order to discuss theoretical identities, which he says are “generally identities involving two rigid designators”(140). He explains this as follows: “Any necessary truth, whether a priori or a posteriori, could not have turned out otherwise. In the case of some necessary a posteriori truths, however, we can say that under appropriate qualitatively identical evidential situations, an appropriate corresponding qualitative statement might have been false”(142).

In virtue of this, he says, we should not say that gold could have turned out to have been a compound, but that it is logically possible that there could have been a compound with all of the properties originally known to hold of gold. This form of possibility is thus defined in relation to our epistemic situation. Instead of saying Hesperus might have turned out not to be Phosphorus, we should say that two distinct bodies could have occupied the position in the sky occupied by Venus.

Kripke tries again to explain the difference between rigid and non-rigid designators, again with the Hesperus-Phosphorus example. He asks us to compare two sentences:

(R) Hesperus = Phosphorus.

(D) The heavenly body in such-and-such position in the sky in the evening = The heavenly body in such-and-such position in the sky in the morning.

The difference, for Kripke, is that R expresses a necessary identity, whereas D expresses a contingent identity. But any argument for R being necessary can apply to D as well.

Kripke’s could point to the fact that “Hesperus" and “Phosphorus" have been baptised to refer to the same object, but the two descriptions in D also fix the same planet as their reference. Kripke could say that these descriptions only contingently refer to the same planet, but that is true of “Hesperus" and “Phosphorus,” as well. We can imagine worlds in which those names were attached to other celestial bodies. Kripke acknowledges that much. He just says that, in those worlds, the other celestial bodies would not be Hesperus and Phosphorus. Okay, but that is because of the same identity relation that is expressed in both R and D.

4. Finally (p. 144), he says, it is time to turn to the identity theory. He begins by distinguishing between token-token identity and type-type identity (mentioning Nagel and Davidson as supporters of the former, and not the latter). I suppose the former would entail that a particular mental state is identical with a particular neurological state, but that types of mental states are not types of neurological states. That seems like an odd idea to me, but Kripke says he is not going to discuss it very much. Instead, he is focusing on type-type identity theory.

Let’s say a person is in a particular neurological (or even, more generally, physiological) state when they experience pain. Kripke says it is “at least logically possible” that they could have been in the exact same physical state but without experiencing pain. When he says “logically possible,” I suppose he means semantically possible? In any case, he acknowledges that this assumption would not be permissible according to identity theory. If pain just is a neurological state, then it is not logically possible that one could exist without the other. Yet, on p. 147, he says some identity theorists claim that the relationship between brain states and sensations is only contingent, and not necessary. He mentions David Armstrong and David Lewis as examples. 

I am not familiar with Armstrong’s work, but let’s look for a moment at what Lewis says. In the text which Kripke references (“An Argument For The Identity Theory,” 1966), Lewis explicitly distinguishes between analytical necessity, which tells us that some causal processes must be identical with mental states, and empirical knowledge, which tells us that “these causal roles belong in fact to certain physical states”(p. 17). In other words, Lewis is saying that the concept of pain, for example, does not contain the concept of neurological states. However, it does contain the concept of physical efficacy, and it is a matter of fact that this is manifest in neurological states. 

According to Kripke’s nomenclature, Lewis is claiming that neurological states are pain states in just the same way that water is H₂0. If Lewis is right, then Kripke should say that, in another world, something other than the specific neurological processes that we experience and call “pain" could be play the role of pain for some creatures; but that is the same as saying that, in another world, something other than H₂0 could fill the water role for other creatures. The fact is that, in our world, “pain" refers to neurological states, and water refers to H₂0.

So how does Kripke represent Lewis?  Well, he doesn’t do a good job of it. He says, "the 'causal role' of the physical state is regarded by the theorists in question as a contingent property of the state, and thus it is supposed to be a contingent property of the state that it is a mental state at all, let alone that it is something as specific as a pain”(147). That is obviously not what Lewis says. Not even close.

But anyway, Kripke acknowledges that it may still be the case that pain is identical to brain states in the same way that water is identical to H₂0. But he is not optimistic that this can be shown. He acknowledges that it is an empirical question, but he says it is an unusual one. He says we normally confuse contingent and necessary facts because we think we can imagine experiencing the same phenomena without the same physical cause. Thus, we (incorrectly, he says) think that we could experience water without experiencing H₂0; and we think we can experience heat without experiencing the transfer of thermal energy. He relies on the notion of “qualitative state” as a kind of epistemic situation here. Because of our epistemic situation, and because of our qualitative experiences, we fail to see the necessary truth of the identity relation. The point, Kripke says, is that our epistemic-phenomenological situation entails a mediation between the experience of what is named and the thing itself. We do not experience heat directly, he says; we experience a sensation which we associate with heat, and we fix the name “heat" to the cause of that sensation. The same with “Hesperus" and so on.

In the case of pain, he says (on p. 151) it is not possible for something being felt as pain without there being any pain. So, he implies, we cannot fix the name “pain" to the cause of the sensation of pain without actually fixing it to pain itself. But this is confused, because pain is not the cause of the sensation of pain. Pain is the sensation of pain. So the issue is not regarding whether we can be wrong about identifying the cause of pain. The issue is whether we can be wrong about naming the sensation itself.

Since the word “pain” is learned, Kripke might have us assume that the referent was fixed before we were even born. Thus, to learn how to use the word “pain,” we must be able to identify the sensation. We might do this by description or through ostensive means. For example, a young child might get hurt and cry, and if they are told that what they are feeling is pain, they will fix the name to the sensation. But they can do this incorrectly. Their feelings can be complex, and they might not pick out only what other people would think of as pain. Over time, as they become more familiar with different sensations, they will come to regard pain in different ways. Their memory of past uses of “pain" will guide them, but these memories can be unreliable. When they apply the word “pain” to a new experience, they might be doing it incorrectly. This new sensation might not be the same as the previous one, so that they are using “pain" to pick out whatever (according to the epistemic situation) they think pain is supposed to be. They could be wrong, in so far as they might not be using the word “pain" as they did before, and as the person did who taught them the word. As an adult, they might learn that it is not always possible to separate pleasure from pain, and they can decide that the term “pain" does not pick out a natural kind at all. 

In all of this, it is evident that an individual learns to use the word “pain" idiosyncratically. There is no empirically fixed body, like a planet or chemical structure, that we can refer to which regulates our use of the word “pain.” It is therefore conceivable that a person can incorrectly think they are using “pain" as a rigid designator for a particular sensation which they undoubtedly feel in a particular moment.

This, I think, is why Wittgenstein argued that to say “I am in pain” is not to state a fact. Rather, saying “I'm in pain” is simply an expression of pain. This is not to say that “I'm in pain” cannot be used to state a fact. But in that case, the situation would be quite different. For example, if a person is reporting a pathological condition in which they regularly feel pain, they can report this fact by saying, “I’m in pain.” They are not reporting a particular pain, but a condition of often experiencing pain.

Let’s consider again how a child learns about pain. Since the word is learned, the referent was fixed before the child ever used it. The child learns to use “pain" in situations where it is a demonstrable fact that the child is in pain. That is, the child is crying, there is an injury, etc. The child is dealing with three phenomena: the naming of a thing, the sensation of an experience, and the physiological demonstration. We want to say that the word “pain" picks out the sensation and not the demonstration. We might say that the physiologically observable facts are circumstantial, that they only provide a convenient opportunity to teach the meaning of the word. But that is not quite right. The particular physiological facts are not necessary for the learning of the word, but there must be some physiologically observable facts. If a child is crying for no apparent reason, a parent might assume they are tired, or hungry. They cannot assume the child is in pain. They might ask the child if they are in pain, but if the child does not know the word, that obviously will not do. They are looking for an opportunity to teach the meaning of the word, and crying by itself is not enough.

We can therefore say that it is analytically necessary that the word “pain" is connected to a physiologically demonstrable situation. This does not stop us from experiencing pain when no such demonstration is possible. We can feel the pain of heartache, and nobody can see that; but we feel it in our chest, and just as with a stomachache, we know that parts of our body are hidden from view. The physiological component of a pain can be circumstantially hidden from view. However, we cannot learn the use of “pain" that way. Therefore, even if nobody can see the physiological signs of heartache, we assume that there is a physiological component. We have learned that pain has physiological signs, and that is how the referent was fixed.

Then we learn that the brain is responsible for all of this, in so far as what we think we feel in particular parts of the body is a matter of stimulations to our nerve fibres. We learn that people can experience phantom pain—pain in parts of the body that no longer exist. So we learn that the sensation of pain is, as a matter of empirical fact, determined by neurological activity.

The analogy to heat and molecules is not so different now. When a person feels pain in their foot and fixes the referent “pain in my foot," we can imagine that they are mistaken. The pain might not be in their foot at all. And in a sense, it is not really in their foot.

Kripke may insist that we fix the name “pain" to a non-localised sensation, to a disembodied phenomenological experience. But that is not correct. As we have seen, pain is always fixed in reference to a physiological fact. We do not experience disembodied pain, because we do not learn to use the word “pain" that way. We fix the word “pain" to embodied experiences.

A doctor asks a patient to tell them when it hurts. The doctor feels around until the patient responds, “It hurts!” Now the doctor concludes that they have found the source of the pain. They believe they have understood the pain now, and they can refer to it as, for example, a bellyache. From the doctor’s point of view, they could have had the same qualitative experience—they could have been in the same epistemic situation—and the person could have not been in pain at all. Perhaps the person lied about the pain. The doctor does not experience the bellyache directly. But, if the person really is in pain, then the doctor really has fixed the referent correctly. In that case, Kripke should say that the doctor’s naming of the pain as a bellyache necessarily corresponds to the pain itself.

Now the doctor is also looking at neurological patterns in the patient’s brain. They identify a particular neurological state and suppose that it might be the pain that the patient is reporting. They suppose that it might not, but let’s say it is. If this is true, then the neurological state is the pain. The doctor thinks it might not be because of their epistemic situation. It is just the same as with water, or heat, or any other object of scientific study.

Kripke argues that pain is not picked out by an accidental property, but by the necessary phenomenological property of being in pain. This is incorrect. The word is learned by picking out accidental properties (injuries, bellyaches) and by remembering those embodied experiences. He says, "Thus it is not possible to say that although pain is necessarily identical with a certain physical state, a certain phenomenon can be picked out in the same way we pick out pain without being correlated with that physical state. If any phenomenon is picked out in exactly the same way that we pick out pain, then that phenomenon is pain”(153).

Kripke’s argument against identity theory only works if there is no mediation between the experience of pain and the fixing of the referent. As I have shown, the experience of disembodied pain is not sufficient to fix the referent. Therefore, there must be mediation. So Kripke’s argument fails.

Kripke now (pp. 153-154) turns to a theological mode of argument, considering an omnipotent creator. I’m not sure how we’re supposed to consider this. Perhaps we can think of God as an omnipotent and omniscience scientist? In any case, Kripke argues as follows: The creation of heat requires only the creation of molecular motion. There is no additional work required to get heat from molecular motion. The reason we need science to recognise that heat is molecular motion is because we experience heat as a particular sensation without understanding the nature of molecular motion. But here we must be careful, because Kripke is saying we use “heat" to name a particular sensation. Is this a disembodied sensation? (Obviously not, because then we could not learn the meaning of the word!) Kripke must say that this sensation is correlated with physically demonstrable facts. Thus we are able to study different physical processes as things which cause us to feel heat. But then we also learn to use the word “cold,” and that is quite a different sensation. Yet cold is also experienced by the transfer of thermal energy from one system to another. So “cold" refers to the same physical processes as “heat.” Should we say that “cold" and “heat" refer to the same thing? From a scientific point of view, perhaps yes. But we do not use them to refer to the same sensations. Are we wrong? Is heat the same as cold in the same way that Hesperus is the same as Phosphorus? Isn’t the only difference in the way we experience them?

Kripke would not say that. The experiences of heat and cold are distinct, and we use the words to refer to those distinct experiences. So the reason we can be mistaken about the relationship between heat and molecular motion is because they are not the same thing.

Maybe the example of Hesperus and Phosphorus would be better for Kripke’s argument. The existence of Venus is sufficient to bring about the existence of Hesperus and Phosphorus, because they are the same thing. The reason we can be wrong about that is because we do not experience Venus directly. We experience it in different circumstances, so that its appearance is not always the same. Thus, we infer the existence of two distinct bodies when there is really only one.

In the case of pain (or heat, for that matter) and neurological states, Kripke says that the creation of the latter should be sufficient for the former (assuming identity theory). We do not experience neurological states in different circumstances, Kripke says, therefore there should be no confusion about their identity.

But this is incorrect. As we’ve seen, pain is felt in wildly different circumstances. We do not fix the name “pain" without attending to accidental circumstances.

Now, let’s step back for a moment. Kripke’s argument is against identifying pain with particular neurological facts. But really, his argument (if sound) would work against identifying pain with any more fundamental facts. If Kripke were right, then pain is an irreducible phenomenon. It cannot ever be understood in other terms. Whereas some substances can be seen as the result of more fundamental processes, pain cannot. No experiential state could ever been understood in terms of more fundamental processes. Thus, the fact that pain is caused by particular physical circumstances would be an intrinsically unsolvable mystery. We could never understand any kind of causal relationship between pain and anything else, because that would require analysing pain in terms of other things. This is impossible, according to Kripke’s argument.

How could we ever fix a referent to something like this? It is certainly not how we learn to use the word “pain.” Indeed, if our parents thought that pain was not causally connected to our physical circumstances, they would not try to teach us the word, would they?

Kripke’s argument does not account for how we actually learn and use the word “pain.” It turns pain into something we could never talk about.

THE END

P.S. Kripke offers additional arguments in the Preface and Addenda, which I may comment on later.