Kirk Rader 1.0-SNAPSHOT
The logic of declarative sentences.
The sentential calculus is a formal language for expressing simple logical arguments. Consider the following:
The first two sentences are premises that are simply stipulated to be true for the purposes of the given argument. The third sentence is the conclusion that is derived from the premises by the application of Inference Rules (a-modus-ponens in this case). The whole argument is a theorem, i.e. it is a valid argument. An argument is valid if and only if it is possible to show that the conclusion must be true when making no assumptions other than that the explicitly cited premises are true and then applying a sequence of one or more transformations that are supported by the axioms of symbolic logic.
The terminology and motivation for symbolic logic predates the modern distinction between a formal system and its meta-mathematical interpretation. It is critical to understand that when logicians pronounce formal expressions like as "if P then Q," they are simply using the common oral terminology for a formal expression involving a Conditional ("If"). It is a meta-mathematical notion to equate such formal expressions with the meaning of the natural language phrases to which they have traditionally been said to correspond. Such meta-mathematical notions are not the domain of discourse for symbolic logic.
Another way of understanding the distinction is by the following observation: any attempt to prove a point about the real world by representing an argument as expressions of symbolic logic must reduce all of the terms of the argument to absolutely unambiguous declarative statements with objectively determined truth values and completely incontestable compositions of those terms using the connectives of symbolic logic as they are defined by axioms of symolic logic and not as they are commonly understood in natural language. Only then can the proof procedures discussed below actually apply to any given argument concerning the state of things in the real world.
Good luck with that!
The preceding natural-language argument could be expressed as a sequence of well-formed formulas* (WFF's) of the sentential calculus, as follows:
The primitive terms in a formula of the sentential calculus are represented by upper case letters of the Latin alphabet, and in this example. They evaluate to one of exactly two possible values, true (T) or false (F). When interpreting an expression in natural language, such primitive terms correspond to complete declarative sentences – hence the "sentential" in "sentential calculus" – such as "It is raining," "Two plus two equals four" and so on.
The symbol is one of a number of connectives that can be used to combine terms to express a logical relationship – "if ... then ..." in this case. The symbol is pronounced "therefore" and is used to mark the proposition that is the conclusion of an argument, while the formulas above the line are its premises.
There are a number of "connectives" or "operators." Each is defined by its truth table* and has a commonly used natural language formulation like "and," "or," "if ... then ..." etc.
Conventionally, greek letters like and are used when documenting axioms and inference rules which apply to any combination of well-formed formulas. E.g. is used when discussing the truth table for or inference rules involving the conditional connective as it could be applied to any combination of well-formed formulas while is a particular conditional formula and is another.
Here, for example, is the truth table for , i.e. "and":
See Connectives for a complete set of such truth tables for all of the connectives in traditional logic and a more detailed explanation of their meaning.
The validity of the first argument, above, should be fairly intuitive. Assuming that the symbol means "not" (logical negation) and the symbol means "or", consider this argument:
This is less immediately intuitive, but is also valid. In fact, according to the rules of the sentential calculus it is entirely equivalent to the first argument involving the symbol, as will be discussed further below. For now, consider the following translation into the corresponding natural language argument:
That the sentence, "Either the weather is fair or the streets are wet," is logically equivalent to the sentence, "If it is raining then the streets are wet," is a theorem of traditional logic – but only assuming that "the weather is fair" means exactly the same thing as "it is not raining" and vice versa. The whole point of symbolic logic as it was originally conceived is to divorce purely logical arguments from that sort of ambiguity of natural language. It is from that point view that we can assert the equivalence:
which could be rendered into English as " \form#59 implies \form#60 if and only if either \form#59 is false or \form#60 is true."
This informal syntax describes the sentential calculus as it is actually written by human beings on white boards and the like. See Appendix 1: Context-Free Grammar for the formal definition of the syntax of the sentential calculus, which is far more strict than described here but which is amenable to automatic proof procedures of the sort that could be carried out by computer algorithms.
The sentence , which may be true (T) or false (F):
Another sentence, , which also may be true or false independently of the value of :
The negation of , i.e. is false if is true, and vice-versa:
Two sentences joined by a connective ("inclusive or", in this case):
Grouping using parentheses:
A vacuously true statement (i.e. a tautology):
A way to visualize the grammar of the sentential calculus is by drawing formulas as tree diagrams. Here is such a tree diagram for the example of De Morgan's Law just cited:
If a node in such a tree corresponds to a primitive term, the truth value of the node is that of the primitive term. Otherwise, the truth value of a node is calculated by first evaluating the truth value(s) of the term(s) to which it is directly connected and then applying the appropriate row from the truth table for the non-terminal node. Applying such a procedure to the preceding diagram shows that the root node is always true without regard to the values of the terminal nodes.