I actually claim to be able to tackle the problem of knowledge in general from the perspective of human action, and therefore, develop a framework for analyzing all methodology. This starts from this analysis, how acting man perceives the world.

Our favorite Robinson Crusoe washes up to his island with no memory. He encounters five fundamental givens, which together enable him to act:

• Sensation,
• Memory,
• Thought,
• Abstract thought¹,
• Unease, and
• Agency.

Sensation is the ultimate given; it is the substance on which thoughts act and of which memory is composed. It refers to particular feelings, in the strict sense: "input" to the brain's program, if one permits a computational analogy tortured all-too-easily. That which they happen to is "I." Memory is a collection of stored sensations not currently happening. Remembering itself is a sensation, and accordingly can be remembered—ordering the memory one remembers having by set containment yields a perception of time. The "I"s of these sensations throughout memory, ordered with respect to time, is the "self." Thought is the process of transforming, combining, manipulating, and producing sensation. It appears to be intentional and dynamical. It is a sensation itself. Abstract thoughts are a class of thoughts whereby the self identifies commonalities between sensations (happening across time, space, memories; across even abstract thoughts themselves) and concieves of that which is common as a sensation in itself, separate (although derived from) the sensations of first order. This can be things like "cup," "predator," "catallactics," or (after a couple thousand years of mathematical progress) "0." Unease is the sensation of there being some sort of grading or ranking on sensations (or, at least, on some of them). Some sensations are preferred to other sensations. Agency is the belief that by thinking certain thoughts (including, but not limited to, thoughts inducing bodily responses) sensations engendering less unease may be made more common.

These conditions are sufficient to meet Mises' prerequisites for action.

They're also necessary for a mind to be capable of action (or, really, to be regarded as a "mind," I should think). For, what if our Crusoe has all but one? Each case leads to Miscesian "contentment or satisfaction." If he lacks sensation, then there is no "he," there is nothing for him to think on, there is nothing to perform action, nothing for his memory to record or his thoughts to abstract from. The case is vacuous. If he lacks memory, then even if he can think, sense, feel unease, and abstract, he is incapable of identifying temporal patterns through the abstract thought (which he may be nevertheless able to apply to information at hand in a particular instant), and accordingly unable to know that there existed an uneasy sensation he might otherwise avoid. As without thought there is no abstract thought, the next case is dispensed with for the same reasons. Without unease, there is no reason for him to prefer one particular sensation to any other, and so no reason for him to purposefully alter anything. Without agency he lacks, as Mises puts it, "the expectation that purposeful behavior has the power to remove or at least to alleviate the felt uneasiness," and so similarly lacks reason to take purposeful action.

So, Crusoe takes action (by assumption). On the basis of his (confused, malformed, naïve) abstractions over sensation, he takes action designed to engender more propitious sensation. And inevitably fails. What is his recourse? Nothing, presuming he considers the entailing sensations to have less unease than those besetting him, other than reevaluating his abstractions and their suitability for his sensations on the basis of his memories regarding the action (those from before, the thoughts leading him to the action, and the sensation after). And so, he comes, through action, to optimize perception: the relation of sensations to abstractions.

Perception always entails these two components. Without sensation, there obviously is no perception (what are you perceiving?). However, without abstractions there also is no perception. With a raw sensation alone, there is no way to decode the information into manageable components. How could I, without abstraction, deduce that the visual stimuli my brain is currently receiving support the sentence "there is a multimeter to the left of my computer monitor?" I must, on the basis of past experience deduce that material objects as such exist, and then that there are such objects called multimeters or computer monitors, that these objects have certain qualities that lead them to look certain ways, that abstract concepts like space, proximity, and direction exist, that "left" is one such direction, that it is that direction and not the other, and so on and so forth. And, importantly, not all sensations may be paired with all abstractions ("multimeter" seems not to be meaningfully related to "Turbo Lover by Judas Priest"). Even if I experience as bare a sensation I can think of, say "pain" with no further qualification, I must have the abstraction of the concept of pain, from past similar experience, to even note it as something deviating from the default.

There are many conclusions one reaches in refining perception through action. Pretty quickly, one must realize that there are material objects that exist independent of your immediate sensing of them (object permanence). Less quickly, one might come to the conclusion that the abstract thoughts one has point to abstract objects independent of human experience (metaphysical platonism, a by-no-means-settled philosophical position). If one acts in the context of a society, one might learn the abstraction of language, and how to map sensations (including abstractions) onto words, or that other people exist whose input is sometimes valuable. These other people have acted in the past, have abstracted, related, and had that perception tested. Perhaps they've borrowed the perception of people still-older. Knowledge or classes of knowlege believed, on the basis of perception, to be inter-contextually or super-contextually useful in action is said to be "true."

I therefore identify three approaches to increasing human knowledge (i.e. improving perception): finding new sensations, finding new abstractions, and improving the relation between the two, all on the basis of existing knowledge, in the interest of perfecting action's capacity to remove unease. The first may be termed "practice," or "experiment;" the second, "mathematics;" the third, "science." ²