The Logic Café
Reference for Logic Review

The ideas described here presuppose that you've already read the tutorials. If not, it's probably best to go back and work your way through those. Then it is recommended that you print this reference and use it for review.

Contents: Section 1: Logic Concepts; Section 2: The Language of Symbolic Logic; Section 3: Quantifiers; Section 4: Basic Symbolization; Section 5: Categorical Logic and Beyond! Section 6: Probability

1. Logic Concepts

Arguments

Think about the following simple example of reasoning.

All living beings deserve respect because life is sacred and the sacred deserves the greatest respect.

What's going on in this sentence? It's not just a claim. Instead, the author is giving reasons for a conclusion (that all living beings deserve respect.) This may well be part of very controversial thinking. But we can better understand the thinking, to reasonably agree or disagree, if we can analyze its non-controversial aspects.

First, we distinguish the reasons (we'll call these "premises") from the conclusion. For clarity, we will sometimes rewrite reasoning in "standard form": writing the premises out first, drawing a line, then writing the conclusion. For the above reasoning about life, the standard form is the following.

Life is sacred.
The sacred deserves the greatest respect.
All living beings deserve respect.

Notice that in the original form the conclusion came before the premises. But standard form reverses this order.

We need a term to apply to reasoning from premises to a conclusion. We will use the word "argument" even though reasoning need not be particularly disputatious:

Definition: An argument is a collection of statements some of which (the premises) are given as reasons for another member of the collection (the conclusion).

Part of this definition involves the notion of a "statement". Statements are true or false: by definition a statement is a sentence which has a "truth value". So, each statement makes a claim, true or false. On the other hand, there are a number of ways in natural language to utter a sentence but not make a statement: one can ask a question, make a request or demand, or utter an exclamation like "Ugh!". But any premise or conclusion of an argument has to have a truth value, so must be a statement.

Distinguishing and Judging Arguments: Validity and Soundness

One of the main points of logic is to be able to distinguish good reasoning from bad. There are two main parts to this process:

1. the judgment of the force or support of premises for conclusion, and
2. the judgment of the correctness of the premises.

The strongest sort of force or support is associated with valid arguments. The idea is that so long as the premises are assumed to be true, the conclusion is inescapable. We make this a bit more precise in the following terms:

An argument is valid just in case it is not possible that its conclusion be false while its premises are all true.
An argument is invalid if and only if it is not valid.

So the definition of validity (the property of being valid) has to do with 1. Our second definition combines judgments 1. and 2.:

An argument is sound if and only if it is both (a) valid and (b) has only true premises.
An argument is unsound if and only if it is not sound.

Think about the following argument. It's very uncontroversial and really rather uninteresting. But that makes it easier to judge.

All whales are mammals.
The animal that played Free Willy is a whale.
The animal that played Free Willy is a mammal.

Notice first that this argument is valid. Even if you don't know anything about whales or Free Willy, it's clear that the conclusion is inescapable given that the two premises (the statements above the line) are true. Second, the premises are true. So, the argument meets the two conditions required for it to be sound.

Now, consider another argument.

All whales live in the Southern Hemisphere.
Shamu (of San Diego, CA) is a whale.
Shamu lives in the Southern Hemisphere.

This argument too is valid. How can you tell? A test is to imagine the premises being true. Here you might have to imagine herding all the whales south of the equator! But imagine it anyway. Then notice that you are automatically imagining the conclusion being true as well. It's impossible for the conclusion to be false while the premises too are true. So, the argument is valid. But, of course, it's not sound. It has a false premise -- imagining that all whales live south of the equator does not make it so.

Now, not all arguments are meant to be valid or sound. We can only give valid and sound arguments when we have the most forceful evidence. When we do argue in this way, the reasoning is deductive; we'll say the study of such reasoning is "deductive logic".

An argument is deductive if and only if its premises are intended to lead to the conclusion in a valid way.

Note the word "intended" that is part of this definition. Whether or not an argument is deductive depends on how it is meant. Often we intend to give a valid argument but fail. (Didn't you ever give a "proof" in geometry class that was meant to validly imply some theorem, only to find you were wrong?) In any case, an argument may count as deductive even when it is not valid; judging an argument as deductive is a matter of interpretation not just logic.

Distinguishing and Judging Arguments: Inductive Reasoning

Frequently we need to give arguments even when our evidence only makes a conclusion likely, but not inescapable. Then our thinking is often called "inductive". For example,

I have surveyed hundreds of students here at ITU and found that less than 10% say they are happy with the new course fees. My sample was selected at random. So, I conclude with confidence that the vast majority of ITU students do not find the course fees acceptable.

Here, the argument's author is clearly claiming that the evidence cited makes the conclusion likely to be true but not a certainty (surveys sometimes do go badly awry, for instance when the participants have some reason to lie.) So, this argument is a clear case of an inductive argument.

An argument is inductive if and only if its premises are intended to lead to its conclusion with high probability.

We do not use the word "valid" for inductive arguments. Rather, an inductive argument whose premises do support its conclusion as intended (i.e., they make the conclusion likely) is called "inductively strong":

An argument is inductively strong if and only if its conclusion is highly probable to be true given its premises.

Inductive strength is a counterpart to validity: by definition, deductive arguments are intended to be valid; inductive arguments are intended to be inductively strong. Of course, people often give arguments falling short of what was intended. That's why we have logic classes! But the point is that "valid" and "inductively strong" play similar roles for deductive and inductive arguments respectively: valid and inductively strong arguments have premises that support their conclusions as intended.

There is an interesting way to state the difference between valid deductive arguments and strong inductive ones. The conclusion of a valid argument is inescapable given it's premises. So, the content of the conclusion is implicit already in the premises. Not so with inductive arguments: their conclusions go beyond the content of their premises. So, inductive reasoning is sometimes called "ampilative" because it amplifies, or adds to, the information given in the premises.

Finally, we need to define a counterpart to "sound" for inductive arguments. Remember, that an argument is sound if and only if it is both valid and has all and only true premises. For an inductive argument, we just substitute "inductively strong" for "valid" to get the notion of cogency:

An argument is cogent if and only if it is both inductively strong and all its premises are true.

So, if one gives an inductive argument, one hopes that it is cogent.

Further Deductive Concepts

There are a couple more logic concepts worth knowing. All these involve possibility and for this reason are associated with deductive logic.

        Quick Overview of the concept of "possibility":

The type of possibility in question in our study of logic is sometimes called logical possibility. Logical possibility is about what might have happened in some possible world, about how things could have been (even if actual matters that have become settled now preclude it).

So, for example, it is logically possible that George W. Bush never became US president. Even though we know that he did, our language allows us to consider a possible, but counterfactual situation in which Gore won in the Supreme Court, votes were recounted and Bush was declared the loser.

There are lots of other uses of the word "possible". You might say: "I'm sure that G.W.Bush didn't lose; it's just not possible that I'm mistaken." This is an epistemic sense of possibility. But our logical possibility is different; it's a semantic conception.

Perhaps the most important deductive concept after validity is that of a logical truth:

A sentence is logically true just in case it is not possible for that sentence to be false.

So, for example, "All Irish males are male" is a logical truth. So, is "Each triangle has three sides".

Sometimes these logical truths are called "analytic" or "necessary truths". But such labels have slightly different definitions and their equivocation with logical truth is controversial. (Only as first approximation should you identify these notions; sorting out the concepts is a very good philosophical exercise.)

Sometimes the notion of necessary truth is given a symbolization:
'S' (read "box-S") symbolizes "it is necessary that S". However, we won't get to this "modal logic" in what follows.

Also, we'll have reason to use:

A sentence is logically false just in case it is not possible for that sentence to be true.

For example, "Agnes will attend law school and it's not the case that she will (ever) attend law school."

Another definition worth keeping in mind is:

The members of a pair of sentences are logically equivalent just in case it is not possible for one to be true while the other is false.

And finally:

One sentence logically implies (or logically entails) a second if and only if it is not possible for the first to be true while the second is false. (In this case we may also say that the second is logically deducible from the first.)

You should think about examples meeting each of these definitions. If you need help, check out the tutorials.

Fallacies

A fallacy is an argument that misleads. It's a "trick" of reasoning. There are two main types of fallacious reasoning: Formal and informal.

Formal Fallacies

These are arguments which are fallacious because of bad form. We've already seen an example of this: the Sanchez case.

Sanchez stays at her banking job only if she gets a raise. So, if she gets a raise, she'll continue at the bank.

This reasoning may at first seem OK. But it's not. To see the problem, notice that the same form of reasoning is obviously wrong in a different context:

There is fire only if there's oxygen. So, if we add oxygen to an area, there will be fire.

Notice that both the last two arguments are of this form:

_____ only if _ _ _ _ _.  So, if _ _ _ _ _ , then ______.

But this is wrong, as the fire-oxygen example shows. When we do symbolic logic, we'll be able to say just what's wrong with these arguments and this form.

Informal Fallacies

1. Ad Hominem Arguments—an argument against the person
These fallacies occur when one arguer attempts to discredit another person rather than his or her argument or position.

Example: If you make an argument claiming to show that God does not exist, and I reply that you're a damned atheist, then my reply is an ad hominem fallacy. It does not show that your argument is unsound or uncogent.

Ad hominem replies usually work by trying to

• attack the character of the arguer
• attack the circumstances of the arguer as indicating a bias
• claim that the arguer is a hypocrite

2. Straw Man—argument by misrepresentation
The arguer attempts to refute another arguer but only by misdescribing his argument or position to make it appear bad, silly, or just weak. The fallacious reasoning misrepresents, making a "straw man" of the opponent.

Example: George W. Bush is an idiot. He thinks he can run a huge government and finance countless wars while cutting taxes to zero for the wealthy.

This is extreme hyperbole. It makes a straw man of Bush's governance ... whatever else you may think of it. If you disagree with the man, good reasoning in support of your position would argue against his actual policies. (Unfortunately, this sort of fallacious argumentation by misrepresentation is some of the most common reasoning in all political discourse. Better to fault Bush for running up the deficit and risking the future tenability of the federal bureaucracy. Perhaps the Bush position is unjust but it's more plausibly crafty than idiotic!)

3. Red Herring—argument by irrelevance

The hunt is on...the fox is released and the dogs are in pursuit; but to make the chase more sporting, a dead and smelly herring is dragged across his path to obscure the scent. Similarly in argument: the red herring fallacy is committed when one arguer tries to obscure the obvious by bringing claims to bear which are only apparently relevant to the issue at hand.

Example: You shouldn't even think about becoming a Catholic priest. Just think about the scandal caused by sexual abuse. We should be very angry.

4. Begging the Question—assuming what needs to be proved
It's all too common for an arguer to slip in – as unnoticed presupposition – a premise that is central to his or her conclusion, in effectg presuming what is just the point of contention!

There are fairly obvious cases of circular reasoning that fit this mold:

Example A: God must exist for the bible tells us so. And we can know the bible is 100% literal truth for it is the result of divine inspiration.

That is: God's existence is "proven" by presupposing that the bible is divine (Godly) inspiration. That's argument in a circle: one type of question begging.

But there are also less obvious cases of begging the question.

Example B: As a good human being you should never eat another mammal. For mammals are all sentient and we should never eat what is sentient.

Here the problem is that the main point of contention, whether or not we should eat sentient creatures, is merely left as a a presupposition and never defended.

(Aside: "begging the question" is now commonly used to mean something different, to mean "the question needs or begs to be asked". We'll ignore this usage here.)

5. Suppressed evidence—leaving out the important part

Example: So, you need to convince a friend to attend DSU with you. You tell her about the great times, the easy grading policies, and the camaraderie the two of you would share. But you conveniently forget to mention the huge cost of tuition and housing.

You've suppressed some of the most relevant information. The conclusion, that she should attend with you, is vastly undermined by the missing information. So, your attempt to convince is a kind of trickery.

2. The Language of Symbolic Logic

Begin by thinking about a simple compound sentence:

(*) Both Jeremy and Karla passed the bar exam, although Jeremy did so before Karla.

In the simplest "sentence" logic, we would represent (*) with something like '(J&K)&B'.

Here the '&' is a connective standing for "and". 'J' stands for "Jeremy passed the bar exam", 'K' for "Karla passed the bar exam", and 'B' for "Jeremy passed before Karla". (NOTICE: We also use '&' for "although" . The idea is that words like "although" mean roughly "and (in contrast)". Words like "although", "but", and "however" mean about the same as "and" but also mark a divergence or distinction.)

But we can do better at the symbolization. Use 'j' and 'k' as names for Jeremy and Karla; then use 'P' to symbolize the predicate "passed the bar exam" and 'B' to symbolize the relationship "passing the bar exam before". We will write 'Pj' for "Jeremy passed the bar exam, 'Pk' for "Karla passed the bar exam" and 'Bjk' for "Jeremy passed before Karla". So, (*) can be symbolized as:

(Pj&Pk)&Bjk

Here we use parentheses to group.

A relationship, like that expressed by 'B' in 'Bjk', is sometimes called a "two place predicate" because it's a predicate relating two things. Can you think of an example of a three place predicate? (More on these in a moment.)

Connectives

English uses many words other than "and" to connect simple sentences to make compound ones.

For example, English uses the word "or" to connect sentences. Our symbolic language uses 'v' instead to express the idea of "either...or...".

So,

Either Karla or Bob passed the bar exam

may be symbolized as

Pk v Pb

English has the words "If...then..." which together connect a pair of sentences. Our symbolic language uses the horseshoe, '>', to express this "conditional". And while the English expresses negation in a number of ways, for exam with "it's not the case", our symbolic language will use just '~'.

Then, for a slightly more complicated example,

If Karla didn't pass, then Bob did.

would be symbolized as

~Pk>Pb

And now we come to a case for which parentheses are important:

It's not true that if Karla didn't pass, then Bob did.

This one needs to be symbolized like so:

~(Pk>Pb)

Again, the parentheses group so that the negation negates the whole "Pk>Pb". This is similar to the English which negates the whole conditional.

Because '>' (like '&', 'v', and '=') connects a pair of sentences, we call it a binary connective. All connectives of SL except one are binary. The one exception is the tilde, '~'. It attaches to a single sentence. For example, '~A'. We call '~' a unary connective.

A synopsis of the third tutorial is presented in the following table. The connectives are listed in the column on the left. Symbols used by different authors can vary as noted. The "component(s)" of a sentence built with a connective is/are just the simpler sentence(s) connected by the connective in question.

	Connective Name	Resulting Sentence Type	Component Names	Typical English Versions	English Statement	Symbolization
&	Ampersand (sometimes a dot or upside down 'v' is used instead)	Conjunction	Conjuncts	"and", "both ... and ... "	Karla and Bob passed the bar exam.	Pk&Pb
>	Horseshoe (sometimes an arrow is used instead)	Conditional	Antecedent, Consequent	"if ... then ... "	If Karla passed, then Bob did.	Pk>Pb
~	Tilde (sometimes a '' is used instead)	Negation	Negate	"it's not the case that", "not"	Bob did not pass the bar exam.	~Pb
v	Wedge	Disjunction	Disjuncts	"or", "either... or... "	Either Karla or Bob passed the bar exam.	Pk v Pb
=	Triple Bar (sometimes a double arrow is used instead)	Biconditional	Bicomponets1	"if and only if", "just in case"	Karla passed the bar exam if and only if Bob did.	Pk=Pb

We are interested in sentences. The atomic ones are those like Bab or Jd or Rmno or just 'K'. Compound sentences, often called molecular sentences, are formed by using connectives. One can use as many connectives as one wishes to "build" a grammatically correct sentence, but the connectives must be added one at a time.

For example, one can use an ampersand, '&', to build 'Bab&Jd' out of the atomic ones. Then go on to make an even longer compound sentence:

(Bab&Jd)>K

Note the parentheses. We are asked always to group when we use binary connectives. Except we are allowed to drop outside parentheses. And we can keep on building, for example we could negate the whole sentence just produced, but to do so we need to return the "dropped" outside parentheses (or use brackets instead):

~[(Bab&Jd)>K]

We can also spell out what sentences formed with a connective mean. For example, any sentence formed with an ampersand, X&Y, is true if and only if both its conjuncts are true. Otherwise it is false. The yellow column below says the same thing. The columns under the other connectives spell out their meaning.

So, we can tell whether a sentence formed from our connectives is true or not just by knowing the truth value of its parts. That is to say that it's truth value is a function of the truth value of its parts. The usual lingo here is that each of the connectives is a "truth function" and that we are developing "truth functional logic".

Summary

We have a language which includes:

names which are lower case letters. So, we can symbolize "Agnes" as 'a' (or any other lower case letter from 'a' through 'u'; 'v' - 'z' are reserved for other uses).

atomic sentences and predicates which are upper case letters. How will we tell the predicates apart from the atomic sentences?

• If an upper case letter is immediately followed by no lower case letters, then it's a sentence letter.
• If an upper case letter is immediately followed by one lower case letters, then it's a one-place predicate.
• If an upper case letter is immediately followed by two lower case letters, then it's a two-place predicate.
• And so on: If an upper case letter is immediately followed by n lower case letters, then it's an n-place predicate.

compound sentences which are constructed out of atomic sentences and connectives. We have to be careful to group with parentheses or brackets whenever we construct a compound sentence with a binary connective. Though we may drop outside parentheses or brackets when we have finished constructing.

A logic with names, predicates and connectives is sometimes called "0th order logic". What's missing in zeroth order logic, what separates it from logic of the first order, is quantification over the objects that are named. In English, words like "all" and "some" serve to quantify. So, we need to add quantifiers to our symbolic language. (A logic with quantification over properties or sets of objects is called second order. We won't get into that for now.)

3. Quantifiers

Thinking about numbers will help us see how to quantify. For example, the English

There is an even number less than three

means that there is at least one thing x, which is even and less than three. In other words:

(*) There is an x, x is even and x is less than three.

Bear with me! There's a reason to go through this example involving the variable 'x' (which you remember using in high school, yes?). Let's begin translating into PL; the above comes to:

(**) There is an x such that: Ex & Lxc

The phrase "there is" indicates a quantifier. It specifies that there is something having certain properties. We will write this with a new symbol, the backward-E: '%'.

(**), then, will be symbolized as follows:

(%x)(Ex&Lxc)

The backward-E is called the "existential" quantifier because it says that something exists.

There is one more quantifier used in our symbolism: the universal quantifier upside-down A: '^'. This quantifier means "all" or "every". We can use it to symbolize the following.

Everyone will attend law school and need a loan

would be:

(^x)(Wx&Nx)

This should be understood to mean:

Everything x is such that it, x, is both W and N.

The use of upside-down A for the universal quantifier is not quite universal! Some write '(x)' instead of '(^x)'.

One last point is in order. When we talk about "something" or "everything" in English, we usually have some particular group of things in mind. For instance, if we say that everyone will attend law school, we don't mean literally everyone in the world. Instead, we may have some circle of friends in mind. Similarly, all quantification assumes a "universe of discourse" the collection of all objects under discussion.

4. Basic Symbolization

Now that we have a sure handle on the syntax (or "grammar") of our new language, we can press forward with semantical issues. The easiest way to do this is to relate PL to English.

Some Symbolization Basics

English has many ways to name an object. The easiest way is the proper name. But there other types of English expressions used to refer to a unique individual. The following English expressions are typically used to signify a specific individual and so can be symbolized with names.

          Names

Proper Nouns like "Paris", "Earth", "Mary", "Oakland University", "Waiting for Godot", "tomorrow", etc.

Kind Names like "oxygen", "Homo Sapiens" (the species), "logic", etc.

Pronouns like: "this", "that", "he", "she", "it", "who", "what", "there", etc.

Definite Descriptions like: "the boy in the field", "Smith's murderer", "the square root of 4", "my son", etc.

Other tags like numerals or symbols, e.g., '(*)' as used in this reference manual.

           Words often symbolized with '%':

"some", "something", "someone", "somewhere", "at least one", "there is", "a", "an", "one"

(Warning: The last three of these fairly often mean something different and not to be symbolized with '%'. For example, "a whale is a mammal" probably means that any whale is a mammal and needs to be symbolized with an '^'.)

It's good to keep some very basic examples in mind:

We may do much the same thing with the universal quantifier.

           Words often symbolized with '^':

"all", "every", "each", "whatever", "whenever", "always", "any", "anyone"

(Warning: The last two of these fairly often mean something different and not to be symbolized with '^'. For example, when I say "if anyone can do it, I can" this may be symbolized as "(%x)Dx>Di".)

Here are some examples.

5. Categorical Logic and Beyond!

It may be best to see languages (like English) as having two basic quantificational forms: the existential and the universal.

Existential Form

The first basic form of English is the following.

existential form:     Some S are P.

where 'S' (the subject) and 'P' (the predicate of the expression) name groups or classes of individuals. (We will call these the subject class and the predicate class, respectively.)

So, for example. "Some students are freshman" is of existential form. And it's pretty easy to see how it might be symbolized. Given a natural symbolization key, it could well be rendered as '(%x)(Sx&Fx)'. For such an easy example, we don't need to think of forms. But for more complicated cases it's best to fit the "mold".

Take this example,

(*) There are female logic students who are juniors set to graduate next year.

Ugh! But we can fit this messy example sentence into the existential form and then symbolize. The following steps will help as you consider such a sentence.

First, here's the mold we need to fit:

(Step I) Some S are P.

Begin by noting that (*) is about "female logic students". So, this is the subject class. And the predicate class, which (*) attributes to its subject is "juniors who will graduate next year".

Now, we need to provide a hybrid English, PL symbolization of the form:

(Step II) (%x)(x is an S  &  x is a P)

For (*) this should be "(%x)(x is a female logic student & x is a junior set to graduate next year)".

Finally, we take this hybrid and restate it in pure PL, something of this form:

(Step III)       (%x)(Sx & Px)

For (*) this means rewriting the subject phrase "x is a female logic student" and the predicate phrase "x is a junior set to graduate next year" into PL. Take this key:

Then the subject phrase becomes: 'Fx&Sxl' and the predicate phrase becomes 'Jx&Gxn'. So, finally we have:

(*)'s Symbolization: (%x)[ (Fx&Sxl) & (Jx&Gxn) ]

Many different English sentences can likewise be seen to fit this form. You may want to review the tutorial for details. In all cases, you move from seeing the English as about a subject and predicate class to a PL symbolization of form (%x)(Sx & Px).

Universal Form

The second form is for sentences saying that all such-and-such are so-and-so. For example, "All Swedes are Europeans". Again we have a subject class and predicate class:

universal form:     All S are P.

Such a universal statement means that anything is such that if it's in the subject class, then it's also in the predicate class. So, our example might be translated as '(^x)(Sx>Ex)'.

In general, we have the same three step process as for existential form. First we need to see that the English sentence is of a form relating a subject to a predicate in the appropriate way:

(Step I) All S are P

Next, we move to the hybrid form:

(Step II)   (^x)( x is an S > x is a P )

Finally we give the symbolization.

(Step III)    (^x)(Sx>Px)

For another example of universal form, think about

(**) All female juniors will graduate next year.

This means:

(Step I) All female juniors are students who will graduate next year.

Notice that the subject is a conjunction. So, we have the hybrid form:

(Step II)   (^x)( x is a female and a junior > x is a student who will graduate next year )

and finally the symbolization:

(Step III) (^x)( (Fx&Jx) > Gxn )

Categorical Logic

Categorical logic treats logical relationships between the types of things (categories) which satisfy one-place predicates. We can use PL to quickly get at the heart of this logic because categorical forms are built from existential and universal form sentences.

Categorical logic recognizes four main types of statement:

Notice from this table that A-form and I-form are (respectively) just what we call "universal" and "existential" forms. The E-form is either universal with negated consequent or negated existential. And the O-form is existential with negated second conjunct.

Now notice that A and O form sentences are "opposites": if one is true, then the other is false. The same relation of opposition holds between E and I forms. We call such pairs contradictories. This fact is represented in the following table:

Complications...

We should see an example of a more sophisticated use of our "1st order logic". Categorical logic is very useful but is nonetheless limited: It's restricted to logical relationships between one-place predicates. We can look at one example that goes beyond categorical logic. Remember:

(*) Both Jeremy and Karla passed the bar exam, but Jeremy did so before Karla.

We last symbolized this as

(Pj&Pk)&Bjk

But we may do better with quantifiers. The idea is that there is an exam, the bar exam, passed first by Jeremy then later by Karla.

(%x)(%y)(%z)[(Ex&Byz)&(Pjxy&Pkxz)]

Or in a logician's English: There is a bar exam x and times y and z with y coming before z such that Jeremy passed bar exam x at time y and Karla passed this exam x at later time z.

Notice that we used this interpretation:

Ex: x is the bar exam; Bxy: time x comes before time y; Pwxy: w passed x at time y. universe of discourse includes times, types of test (including the bar exam) and people.

Mathematical Logic

We can move on from PL to give a logic sufficient for mathematics with a couple of additions. First we need to add an identity relation for '='. It's natural to pick 'I'. The only difference between 'I' and all other relations is that we also give rules of inference for how 'I' is used. Also we need to add functions! Ugh? Well, just think of functions as complex names. In English we might say "the youngest brother of person p". This is a function from people to other people. For the details, see chapter nine from the Logic Café.

6. Probability

Probability plays a role in inductive logic that is analogous to the role played by possibility in deductive logic. For example, a valid deductive argument has premises which, granted as true, make it impossible for the conclusion to be false. Similarly, a strong inductive argument has premises which, granted as true, make it improbable that the conclusion is false.

The tutorials contain the briefest of introductions to the interpretation of a theory of probability. Here we only give the axiomatic theory.

We'll just take probability as applying to sentences of our symbolic language. For example, we'll write 'P[Wa]' to stand for say "the probability that Agnes will attend law school". Or, 'P[(%x)Wx]' for the probability that someone will attend law school. For our purposes, we'll restrict our new formal language to include PL and any PL sentence surrounded by 'P[...]'.

We will need 5 basic "axioms" of probability:

1.   0 < P[X] < 1

2.   If X is a logical truth, then P[X]=1.

3.   If X and Y are logically equivalent, then P[X] = P[Y].

4.   P[~X] = 1 - P[X]

5.   P[XvY] = P[X] + P[Y] - P[X&Y]

Conditional Probability and Independence

We often describe probabilities in less absolute terms. Instead of saying that your probability of passing this class is high, I say something like "you have a very high probability of passing given that you continue your good work".

That is, we put a condition on the probability assignment. We'll write the probability of X given Y as 'P[X|Y]' and define it this way:

Definition 1:   P[X|Y] = P[X&Y] / P[Y]

Finally, consider:

Definition 2: X and Y are independent if P[X|Y] = P[X]

Bayes' Theorem

Ugh? But this one takes just a little work to prove. And it's worth it. Think about X as a hypothesis and Y as the evidence. Then the left hand side of Bayes' theorem gives the probability of the hypothesis given the evidence. Just what we'd like to be able to know! And the right had side provides the answer partly in terms of how hypotheses provide probabilities for experimental results (evidence). Something we might know. Here's a simplified version of the theorem (with 'H' for the hypothesis and 'E' for the evidence):

Thus we have the basis for an epistemology of science: Bayesian Epistemology.