Behaviorism held sway in psychology until the 1960’s. The more extreme forms of behaviorism were philosophically on shaky ground from the start. For example, some behaviorists denied the existence of mental events altogether. Philosophers generally found that position a little extreme: it seems obvious that even if I can’t be sure anybody else has mental events, I do, and so I can’t deny that mental events exist. (Maybe I’m the only one who has them, but that’s a different problem. Maybe there’s no continuous me to have them, but that’s yet another problem.) Besides, the behaviorists often didn’t really eliminate intentional language at all; they simply substituted less intentional sounding words, e.g., "drive" for "instinct," "food-seeking behavior" for "hunger," "flight behavior" for "fear," etc.

Gilbert Ryle (1900-1976), a British philosopher of the "ordinary language" school, tried to remedy these problems. Like the behaviorists, Ryle opposed the Cartesian view. But Ryle, like Aristotle, opted for metaphysical pluralism. Ryle thought there were many ways to be, and that it was silly to claim that something had simply to be either "mental" on the one hand or "physical" on the other with no merging and no in-between. Ryle saw that if you limited yourself to the Cartesian categories, you ended up making category mistakes (like saying, if you’re a materialist, that intentions can’t exist because they’re "mental"; or if you’re a dualist, that intentional behavior can’t be observed because the intention is "mental"). The root problem is the Cartesian notion of the "mental" as opposed to the "physical." Ryle came up with a new form of behaviorism, which he called logical or philosophical behaviorism. Logical behaviorism was opposed to methodological behaviorism, which was behaviorist psychology’s policy never to use mental events in its "scientific" explanations of behavior. 

So Ryle did not try to deny the existence of mental events or intentionality; on the contrary, Ryle (following Brentano) said you must use intentional language  to understand the behavior of people and higher animals. But according to Ryle, you can take any claim about "the mental" and systematically reformulate it as a set of "if … then" statements (conditionals), where all components (the antecedents and consequents) are observable behaviors. This way you can accept and use mentalistic and intentional language without committing yourself to all the Cartesian spook baggage. Everything that is is publicly; there is one world, for epistemological purposes. But there are many ways to be in that world (metaphysical pluralism). 

Logical behaviorism and non-intentional states (e.g., sensation)

Let’s look at how Ryle’s logical behaviorism would reformulate a statement about a  non-intentional "mental" state such as sensation. Let’s consider the sensation of pain.  According to logical behaviorism, the statement "John is in pain" MEANS the entire set  of true conditional statements such as the following:

The clever idea here is that if our list of conditionals is exhaustive enough, we’ll know  exactly what it MEANS to say "John is in pain" without having to acknowledge any spooky  invisible feelings, since in all these restatements, the referents of both the "if" clause  (the antecedent) and the "then" clause (the consequent) are public and observable.

Next let’s consider how Ryle would reformulate intentional statements such as "Jill believes it is going to rain". For Ryle statements about belief are statements about dispositions to behave. Those dispositions to behave are just the true conditional statements such as the following:

IF		THEN
Jill leaves home		Jill takes her umbrella.
Jill’s windows are open		Jill closes the windows.
Jill’s garden tools are out		Jill puts the tools in the shed.
Jill is wearing her good shoes		Jill changes her shoes.
Jill drives her car		Jill makes sure her windshield wipers are working.
And so on ...

Again, the clever idea here is that in all these restatements, the referents of both the "if" clause (the antecedent) and the "then" clause (the consequent) are public and observable, so we know what it MEANS for Jill to believe it’s going to rain without ever bringing in mysterious subjective private mental events called beliefs.

PROBLEMS:

1. While logical behaviorism is philosophically an improvement over classic behaviorism, there seems no room in logical behaviorism’s ontology for "raw feels" (aka qualia). And without qualia, logical behaviorism can’t answer the Superactor objection: what if someone were such a perfect actor that she could simulate pain behavior in absolutely all respects but felt nothing? Conversely, what if someone were such a perfect Spartan that he could feel terrible pain without displaying any behavioral signs? Logical behaviorism would have to say the SuperActor was in pain and that SuperSpartan wasn’t, and both conclusions would be false. Thus, there’s more to the sensation of pain than a set of conditionals.
2. You need to know everything about someone’s worldview and projects in order to predict that person’s beliefs. In order to formulate the conditionals, you need complete information about everything that stands for everything else within a person’s world view — what someone sees things as. And there’s just too much of that stuff. For examples, see separate notes on  Davidson.
3. If all we mean by dispositions are these conditionals, and these conditionals make no claims about causality, then dispositions have no causal power. Logical behaviorism is nominalistic, in the medieval sense: there are only specific conditional linkages, and nothing more. Since the theory makes no causal claims, it is useless for science.