Financial Mathematics Text

Monday, December 17, 2012

That's an Empirical Question: Part III

This is a follow-up in a series exploring what the difference between "empirical" versus "mathematical" versus "philosophical" questions. For the other posts in the series see:

That's an Empirical Question: Part I
That's an Empirical Question: Part II

I'd like to consider a thought experiment though I think you can probably find some similar at manufacturing firms in the real world.

Suppose we have a machine that produces widgets. These widgets need to be 1 cm tall +/- 1 mm (e.g. 0.9 cm to 1.1 cm). The machine has an measuring device built into it that measures each widget. If the widget is to specifications it gets sorted to the "good" pile. If it is not to specifications, it gets sorted to a "reject" pile.

Each measurement is coded as data by the machine. This data can be accessed by the machine operator. Various statistical calculations can be performed on the data (mean, median, standard deviation, range, % of total parts produced rejected, etc).

Two questions can arise:
  1. Is this set of data obtained on the height of these widgets "empirical"?
  2. In what sense are these data "derived from experience"?
There's a tradition in philosophy that I would like to call Garbage Empiricism.  To oversimplify it, they hold to belief that all knowledge is ultimately derived from something like "sense impressions" (Hume) or "sense data" (Ayer).

If I were a Garbage Empiricist answering the above questions, I see two possible responses that would be consistent:
  1. I would either have to answer "yes" to the first question and commit myself to the belief that the machine taking measurements and storing data is somehow "derived from experience". I'm sure the mental gymnastics that would be required to make the latter claim would be entertaining on some level.
  2. Alternatively I could answer "no" to the first question. If that's the case, there would be no required mental gymnastics to resolve the second question. But now I'm contending that the data I obtained was non-empirical and therefore would have to conclude that either (1) all knowledge is not empirical or (2) the data obtained wasn't knowledge.
Luckily for me I'm not a Garbage Empiricist. So here are some thoughts regarding those two questions.

For starters, I suspect that most would consider the height of the widgets to be an "empirical matter". We could ask the empirical philosophers to do some surveys to confirm whether or not this is a generally held belief or not.

As a result, I would conclude that "experience" or "sense data" is not an essential requirement for "empiricism". Perhaps I am describing something different from empiricism here (again, we haven't defined any terms as of yet.) But I would contend that measurement is the primary "raw material" from which we form knowledge.

In physics, which is often regarded as the pinnacle of the "empirical" sciences, we often speak of observations. An observation is, in most cases, a measurement, not some "mental act" by some "conscious agent". Measurements involve interacting with that which we are measuring. This tangent is probably worth for more than the treatment I can give here. But I'd like to offer a general outline for what I consider good empiricism:

Hypothesis for Empirical Knowledge/Questions

Good empiricism requires observations.

Observations requires some form of interaction with that which is being observed.

It is this interaction that makes a particular piece of knowledge "empirical".

This is just a brief outline of a working hypothesis. But I believe it can offer some insight.

In the case of my thought experiment, the measurements obtained on the height of the widgets is "empirical" because the act of measuring was a form of interaction with the widgets that obtained specific information about it. Whether or not "sense experience" was involved would be inconsequential for satisfying the above definition.

If you're interested in this series, see Part IV.

No comments:

Post a Comment

Some common OpenID URLs (no change to URL required):