19 ideas
| 10751 | Second-order logic needs the sets, and its consequence has epistemological problems [Rossberg] |
| Full Idea: Second-order logic raises doubts because of its ontological commitment to the set-theoretic hierarchy, and the allegedly problematic epistemic status of the second-order consequence relation. | |
| From: Marcus Rossberg (First-order Logic, 2nd-order, Completeness [2004], §1) | |
| A reaction: The 'epistemic' problem is whether you can know the truths, given that the logic is incomplete, and so they cannot all be proved. Rossberg defends second-order logic against the second problem. A third problem is that it may be mathematics. |
| 10757 | Henkin semantics has a second domain of predicates and relations (in upper case) [Rossberg] |
| Full Idea: Henkin semantics (for second-order logic) specifies a second domain of predicates and relations for the upper case constants and variables. | |
| From: Marcus Rossberg (First-order Logic, 2nd-order, Completeness [2004], §3) | |
| A reaction: This second domain is restricted to predicates and relations which are actually instantiated in the model. Second-order logic is complete with this semantics. Cf. Idea 10756. |
| 10759 | There are at least seven possible systems of semantics for second-order logic [Rossberg] |
| Full Idea: In addition to standard and Henkin semantics for second-order logic, one might also employ substitutional or game-theoretical or topological semantics, or Boolos's plural interpretation, or even a semantics inspired by Lesniewski. | |
| From: Marcus Rossberg (First-order Logic, 2nd-order, Completeness [2004], §3) | |
| A reaction: This is helpful in seeing the full picture of what is going on in these logical systems. |
| 10753 | Logical consequence is intuitively semantic, and captured by model theory [Rossberg] |
| Full Idea: Logical consequence is intuitively taken to be a semantic notion, ...and it is therefore the formal semantics, i.e. the model theory, that captures logical consequence. | |
| From: Marcus Rossberg (First-order Logic, 2nd-order, Completeness [2004], §2) | |
| A reaction: If you come at the issue from normal speech, this seems right, but if you start thinking about the necessity of logical consequence, that formal rules and proof-theory seem to be the foundation. |
| 10752 | Γ |- S says S can be deduced from Γ; Γ |= S says a good model for Γ makes S true [Rossberg] |
| Full Idea: Deductive consequence, written Γ|-S, is loosely read as 'the sentence S can be deduced from the sentences Γ', and semantic consequence Γ|=S says 'all models that make Γ true make S true as well'. | |
| From: Marcus Rossberg (First-order Logic, 2nd-order, Completeness [2004], §2) | |
| A reaction: We might read |= as 'true in the same model as'. What is the relation, though, between the LHS and the RHS? They seem to be mutually related to some model, but not directly to one another. |
| 10754 | In proof-theory, logical form is shown by the logical constants [Rossberg] |
| Full Idea: A proof-theorist could insist that the logical form of a sentence is exhibited by the logical constants that it contains. | |
| From: Marcus Rossberg (First-order Logic, 2nd-order, Completeness [2004], §2) | |
| A reaction: You have to first get to the formal logical constants, rather than the natural language ones. E.g. what is the truth table for 'but'? There is also the matter of the quantifiers and the domain, and distinguishing real objects and predicates from bogus. |
| 10756 | A model is a domain, and an interpretation assigning objects, predicates, relations etc. [Rossberg] |
| Full Idea: A standard model is a set of objects called the 'domain', and an interpretation function, assigning objects in the domain to names, subsets to predicate letters, subsets of the Cartesian product of the domain with itself to binary relation symbols etc. | |
| From: Marcus Rossberg (First-order Logic, 2nd-order, Completeness [2004], §3) | |
| A reaction: The model actually specifies which objects have which predicates, and which objects are in which relations. Tarski's account of truth in terms of 'satisfaction' seems to be just a description of those pre-decided facts. |
| 10758 | If models of a mathematical theory are all isomorphic, it is 'categorical', with essentially one model [Rossberg] |
| Full Idea: A mathematical theory is 'categorical' if, and only if, all of its models are isomorphic. Such a theory then essentially has just one model, the standard one. | |
| From: Marcus Rossberg (First-order Logic, 2nd-order, Completeness [2004], §3) | |
| A reaction: So the term 'categorical' is gradually replacing the much-used phrase 'up to isomorphism'. |
| 10761 | Completeness can always be achieved by cunning model-design [Rossberg] |
| Full Idea: All that should be required to get a semantics relative to which a given deductive system is complete is a sufficiently cunning model-theorist. | |
| From: Marcus Rossberg (First-order Logic, 2nd-order, Completeness [2004], §5) |
| 10755 | A deductive system is only incomplete with respect to a formal semantics [Rossberg] |
| Full Idea: No deductive system is semantically incomplete in and of itself; rather a deductive system is incomplete with respect to a specified formal semantics. | |
| From: Marcus Rossberg (First-order Logic, 2nd-order, Completeness [2004], §3) | |
| A reaction: This important point indicates that a system might be complete with one semantics and incomplete with another. E.g. second-order logic can be made complete by employing a 'Henkin semantics'. |
| 15034 | Are genera and species real or conceptual? bodies or incorporeal? in sensibles or separate from them? [Porphyry] |
| Full Idea: I shall beg off talking of a) whether genera and species are real or situated in bare thoughts alone, b) whether as real they are bodies or incorporeals, and c) whether they are separated or in sensibles and have their reality in connection with them. | |
| From: Porphyry (Isagoge ('Introduction') [c.295], (2)) | |
| A reaction: This passage, picking up on Aristotle, seems to be the original source that grew into the medievel debate about universals. It seems to rather neatly lay out the agenda for the universals debate which is still with us. |
| 8915 | How we refer to abstractions is much less clear than how we refer to other things [Rosen] |
| Full Idea: It is unclear how we manage to refer determinately to abstract entities in a sense in which it is not unclear how we manage to refer determinately to other things. | |
| From: Gideon Rosen (Abstract Objects [2001], 'Way of Ex') | |
| A reaction: This is where problems of abstraction overlap with problems about reference in language. Can we have a 'baptism' account of each abstraction (even very large numbers)? Will descriptions do it? Do abstractions collapse into particulars when we refer? |
| 8917 | The Way of Abstraction used to say an abstraction is an idea that was formed by abstracting [Rosen] |
| Full Idea: The simplest version of the Way of Abstraction would be to say that an object is abstract if it is a referent of an idea that was formed by abstraction, but this is wedded to an outmoded philosophy of mind. | |
| From: Gideon Rosen (Abstract Objects [2001], 'Way of Abs') | |
| A reaction: This presumably refers to Locke, who wields the highly ambiguous term 'idea'. But if we sort out that ambiguity (by using modern talk of mental events, concepts and content?) we might reclaim the view. But do we have a 'genetic fallacy' here? |
| 8912 | Nowadays abstractions are defined as non-spatial, causally inert things [Rosen] |
| Full Idea: If any characterization of the abstract deserves to be regarded as the modern standard one, it is this: an abstract entity is a non-spatial (or non-spatiotemporal) causally inert thing. This view presents a number of perplexities... | |
| From: Gideon Rosen (Abstract Objects [2001], 'Non-spat') | |
| A reaction: As indicated in other ideas, the problem is that some abstractions do seem to be located somewhere in space-time, and to have come into existence, and to pass away. I like 'to exist is to have causal powers'. See Ideas 5992 and 8300. |
| 8913 | Chess may be abstract, but it has existed in specific space and time [Rosen] |
| Full Idea: The natural view of chess is not that it is a non-spatiotemporal mathematical object, but that it was invented at a certain time and place, that it has changed over the years, and so on. | |
| From: Gideon Rosen (Abstract Objects [2001], 'Non-spat') | |
| A reaction: This strikes me as being undeniable, and being an incredibly important point. Logicians seem to want to subsume things like games into the highly abstract world of logic and numbers. In fact the direction of explanation should be reversed. |
| 8914 | Sets are said to be abstract and non-spatial, but a set of books can be on a shelf [Rosen] |
| Full Idea: It is thought that sets are abstract, abstract objects do not exist in space, so sets must not exist in space. But it is not unnatural to say that a set of books is located on a certain shelf in the library. | |
| From: Gideon Rosen (Abstract Objects [2001], 'Non-spat') | |
| A reaction: The arguments against non-spatiality of abstractions seem to me to be conclusive. Not being able to assign a location to the cosine function is on a par with not knowing where my thoughts are located in my brain. |
| 8916 | Conflating abstractions with either sets or universals is a big claim, needing a big defence [Rosen] |
| Full Idea: The Way of Conflation account of abstractions (identifying them sets or with universals) is now relatively rare. The claim sets or universals are the only abstract objects would amount to a substantive metaphysical thesis, in need of defence. | |
| From: Gideon Rosen (Abstract Objects [2001], 'Way of Con') | |
| A reaction: If you produce a concept like 'mammal' by psychological abstraction, you do seem to end up with a set of things with shared properties, so this approach is not silly. I can't think of any examples of abstractions which are not sets or universals. |
| 8918 | Functional terms can pick out abstractions by asserting an equivalence relation [Rosen] |
| Full Idea: On Frege's suggestion, functional terms that pick out abstract expressions (such as 'direction' or 'equinumeral') have a typical form of f(a) = f(b) iff aRb, where R is an equivalence relation, a relation which is reflexive, symmetric and transitive. | |
| From: Gideon Rosen (Abstract Objects [2001], 'Way of Abs') | |
| A reaction: [Wright and Hale are credited with the details] This has become the modern orthodoxy among the logically-minded. Examples of R are 'parallel' or 'just as many as'. It picks out an 'aspect', which isn't far from the old view. |
| 8919 | Abstraction by equivalence relationships might prove that a train is an abstract entity [Rosen] |
| Full Idea: It seems possible to define a train in terms of its carriages and the connection relationship, which would meet the equivalence account of abstraction, but demonstrate that trains are actually abstract. | |
| From: Gideon Rosen (Abstract Objects [2001], 'Way of Abs') | |
| A reaction: [Compressed. See article for more detail] A tricky example, but a suggestive line of criticism. If you find two physical objects which relate to one another reflexively, symmetrically and transitively, they may turn out to be abstract. |