Modal and counterfactual reasoning is fundamental to most kinds of practical reasoning and in fact to language understanding in general. When, we plan, predict or simply interpret facts in a context of limited knowledge, we consider possible situations (i.e. worlds) that fit with what we know/believe and infer what is necessary and/or possible given this limited knowledge. Indeed, the classical notion of validity can itself be regarded as modal, because it is based on considering what conclusions must hold in all possible situations in which certain premisses hold. More generally, the meaning of concepts does not depend only on characteristics of the actual world but also on how they would be correctly applied in other situations which may or may not occur. Modal logic just makes more explicit, what is already implicit in language use and reasoning. When we say something like `If you jump off that cliff, you will die' we are doing counterfactual reasonining, not just truth-functional reasoning about the actual world (i.e. `if' is not often interpreted as simply a material conditional). I could go on, but I think I have made my point.

Modal and counterfactual reasoning is fundamental to most kinds of practical reasoning and in fact to language understanding in general. When, we plan, predict or simply interpret facts in a context of limited knowledge, we consider possible situations (i.e. worlds) that fit with what we know/believe and infer what is necessary and/or possible given this limited knowledge. Indeed, the classical notion of validity can itself be regarded as modal, because it is based on considering what conclusions must hold in all possible situations in which certain premisses hold. More generally, the meaning of concepts does not depend only on characteristics of the actual world but also on how they would be correctly applied in other situations which may or may not occur. Modal logic just makes more explicit, what is already implicit in language use and reasoning. When we say something like `If you jump off that cliff, you will die' we are doing counterfactual reasonining, not just truth-functional reasoning about the actual world (i.e. `if' is not often interpreted as simply a material conditional). I could go on, but I think I have made my point.

This is a good response Brandon. I agree with what you've said here. I do however get exercised at how much weight is sometimes applied to certain types of modal thought, and I feel that some philosophers try to make it do more than it is capable of doing, but your statement is a good declaration of what it does well. You sound well-informed in this regard so I will unpack my whole concern to you.

Twin earth discussions are, by definition, analogical and not actual. They are thought experiments. Thus, it is important to remember that at base what Horgan and Timmons were attacking in Richard Boyd’s work when they wrote New Wave Moral Realism Meets Moral Twin Earth, was Boyd’s understanding of what a referent actually does. As regards referents, the Kripkean notion of them is clear and straightforward, however, although Putnam’s “water/ twater” discussion harmonizes well with Kripke’s understanding, it ties the notion of what a referent is to the modal device of Twin Earth discussions. Horgan and Timmons appropriated Putnam’s modal device and used it in an ethical direction. It appears in their work that the basic discussion is really over the contrast between Boyd’s primarily epistemic concept of a referent and Kripke’s idea of “ostension and baptism,” which is essentially a descriptive theory of naming. That is where the real issue lies. Therefore, in my opinion, it is doubtful that one can shift the burden of proof from one party to another in this debate simply by means of Twin Earth reasoning. In order to attempt to do so, there are, what I regard as questionable ideas set forth. Dreier has stated my position on this quite well in a response given elsewhere to Blackburn:

"It is a part of the meaning of the word "good" that it names a natural property, just as it is part of the meaning of the word "water" that it names a physical kind. Anyone who thinks two things could be alike in their physical properties while differing in whether they are made of water, reveals ignorance of the meaning of "water". That is a fact about our linguistic practices. And anyone who thinks two things could be alike in their natural properties while differing in their moral goodness, reveals ignorance of the meaning of "good". That is a fact about our linguistic practices. But being made of water and being good are real properties nonetheless."

I believe that Horgan and Timmons are committed to a reductionist form of physicalism. I have a question here on RG that addresses supervenience, and it ties into this whole discussion. In light of what you said simply and straightforwardly here I suppose my problem isn't so much with modal argumentation, but with trying to make it do too much.

It is certainly true that vocabulary derived from modal logic has often been enlisted to rhetorically support arguments that go far beyond what follows from the actual semantics and proof theory of modal logic(s). As I said in my earlier answer, modal logics do not take us much beyond pure classical logic, although they do bring into the object language certain semantic operators that would otherwise only be apprehended via meta-level reasoning. There are some examples of analysis using modal logic that that do yield results of genuine philosophical interest. For instance if we are interested in logical dependencies between different modalities, or whether certain modalities are independent, many interesting results are to be had. An example would be the pure modal derivation of the so called Fitch's 'paradox', which shows by reductio ad absurdum, from very general modal principles, that there must be facts that are true but unknowable.

Many so called 'Twin Earth' arguments seem to me to be deeply flawed. The problem is that we cannot simply postulate a possible world that differs from the actual one with respect to a particular fact or delimited set of facts. This is because there is a very high degree of interdependence between facts that describe the world, both in terms of physical laws and in terms of semantic interdependence between concepts (as well as purely logical dependencies). If we postulate a world in which there is a substance that has the same properties as water but is chemically distinct from water, this will entail radical changes not only in the atomic or sub-atomic constituents of that world but also in the chemical and hence the physical laws that govern that world.

It must be remembered that the 'possible worlds' of modal logic correspond to classical 'models' not to arbitrary sets of facts. Even if we allow possible worlds incorporating arbitrary physical constituents and physical laws, the physical configuration of each possible world must be consistent with its physical laws. This already places very strong constraints on the set of possible worlds and the sets of propositions that could be true in some possible world. If we also want to consider propositions that describe the world in terms of high-level abstract concepts, a further layer of constraints come into play: semantic dependencies between concepts and conditions linking high level vocabulary to low level physical properties of the world. Thus, in order to place a 'Twin Earth' type argument on solid ground, in terms of modal logic, a huge amount of foundational analysis needs to be done to construct a conceptual framework (an ontology) within which the set of possible words can be rigorously defined.

You have articulated the primary difficulties with Twin Earth arguments quite thoroughly and concisely. I believe that Horgan and Timmon's response to Boyd is lacking the necessary foundational work which you speak of here. It has been a catalyst for a whole new trajectory of discussion and debate which I think it simply does not warrant, due to its basic lack of substance and depth.

A science field in which counterfactual reasoning is the by far most used way of thinking is Epidemiology, the basic idea is to select two populations 'differing only' by a given factor of which we want to prove the biological effect (being it a drug or a toxic agent) and, once demonstrated the scoring of a relevant difference between the two populations as for the end-point of interest (morbidity, mortality, recovery after a disease..) the causal link between the agent and the biological end-point is taken for granted. This way of thinking scored incredible successes when the agent of interest had a very strong causal link with the end-point and was mainly 'external' to the biological and environmental features, this was the case of infectious diseases where the pathogenesis was strongly related to 'external' agents like viruses or bacteria but was much less effective for more 'internal' causes as genes, ethnicity, sex...

Let's imagine to design an epidemiological study to check the effect of ethnicity on the prevalence of a multifactorial disease, let's say diabetes, if we want to design a counterfactual test bench we should imagine two 'white' and 'black' populations having the same income, the same food habits, the same education, living in the same place, having the same age structure...going more and more in depth the only possibility is to take a part of the white population and ask them to piant there face in black (or viceversa to ask some memebrs of the black population to paint their face white). Practically the counterfactual approach is a nightmare for this reason a large use of covariate analysis is made adjusting for age effect, food effect, education effect...but this is not satisfactory at all because we must impose a linear and additive combination of effects (plus very basic paired interactions) ..this is why for example the number of 'proven human carcinogens' is incredibly short and the GWAS (genome wide association studies) aimed at isolating 'disease related' genes was a total failure.

In my opinion we should revert this kind of approach to a more ecological one looking exactly for the opposite: instead of trying and isolate a purely ideal (but not feasible in practice) counterfactual apparatus is better looking a posteriori to the invariance of some basic correlations in widely different general contexts (like in meta-analysis).

Thank you Alessandro - this is very interesting and informative. Picking up on your illustration, what would this type of analysis look like if (assuming we weren't already aware of its nature) we were trying to identify the etiology of a disease like sickle-cell anemia? I ask this because it, quite literally, already has race as a primary known factor. If you were an epidemiologist trying to prove your case for a racial basis for sickle-cell anemia, what sort of a counterfactual test might you devise? I realize you said it works better for external factors such as bacterial agents, etc., but what I am curious about is how does one construct an epidemiological counterfactual model in the first place?

Well, sickle-cell-anemia has a very strong genetic primary cause :Sickle-cell disease (SCD), or sickle-cell anaemia (SCA) or drepanocytosis, is a hereditary blood disorder, characterized by red blood cells that assume an abnormal, rigid, sickle shape. Sickling decreases the cells' flexibility and results in a risk of various complications. The sickling occurs because of a mutation in the haemoglobin gene. Individuals with one copy of the defunct gene display both normal and abnormal haemoglobin. Thus we are in a very specific situation in which the 'felon' is very evident: mutated hemoglobi. By the way hemoglobin is in an almost unique position in the realm of biology, being a SINGLE PROTEIN THAT WORKS PRACTICALLY IN ISOLATION with negligible interactions with other proteins (in the 99.9% of cases proteins work in network with other proteins) and IMMEDIATELY EXPOSED to internal milieu (the efficiency of gaseous exchanges is readily explained by structural modifications of the protein).

Now, the sickle cell mutated gene (that acts as co-dominant, so having a 'wild' and a 'mutated' copy drastically lowers the gravity of the symptomps) gives a big advantage as for resistance to malaria and, being practically safe when in heterozigosis, is much more diffused in tropical and sub tropical zones than in other zones of the globe. This provokes a disproportionate frequency of the mutation in black people for the simple fact in Africa malaria is endemic, thus the racial origin of the disease is a 'fake' or, if you prefer, a contingency due to the association between malaria and tropics, in a strictly 'counterfactual' design aimed at looking at 'really causative' agents we should discard the racial basis of the disease (even if in the black population is much more diffuse) because it comes from a confounding effect due to the advantage given by the mutated gene to people exposed to malaria. This is a very good example of the problems coming from this approach...

Thank you Alessandro - that is fascinating!

Grazie a te

Sei il benvenuto.