Ought and Is

Experts are typically expert in two types of things:

  • what is the case
  • what ought to be the case


And for that matter, we all have attitudes and beliefs regarding both types of statement. For example:

  • it's 20 degrees in here
  • it ought to be warmer

That's fair enough.

Science and research are sometimes depicted as consisting necessarily only of the first sort of statement: that is, researchers should confine themselves to discussing what is, and not what ought to be the case.

I don't agree with this. I think that unless research is guided by some sense of ought there is no motive nor even mechanism for determining what is. For example, if it didn't matter what the temperature is, why would anyone bother measuring it, and indeed, what sort of scale would we even use?

Indeed, it's pretty hard to make sense of any human endeavour without invoking both an is and an ought. Any action plan, indeed, contains these two elements: the action plan is a description of the route from the is to the ought. And all research is based on this formulation.

Having said all of this, there are two hard and fast rules that apply regarding inferences involving is and ought:

1. You cannot derive an ought from an is

Suppose it's 20 degrees in here. Does it follow that it should be warmer or colder? No. It depends on your point of view. If you're a duck, you're probably OK with 20 degrees. But if you're me, you want it warmer. And if you're a rock, it doesn't matter at all.

It's sometimes hard to see this. "Look at that starving child," someone will say. Well, yes, the child is starving. But that by itself doesn't allow us to infer that the child should be fed. The inference follows only if we have an expression of need or value, for example, "allowing children to starve is wrong."

This is sometimes called the 'naturalistic fallacy'. People say, for example, "It's human nature to do such-and-such, so such-and-such is OK." Or, conversely, they say something is wrong because "it's not natural."


2. You cannot derive an is from an ought


"If wishes were horses," goes the old saying, "then beggars could ride." There's wisdom in that. Certainly we may believe things ought to be one way or another. But this belief doesn't mean that anything actually is one way or another. This would be nothing more than wishful thinking.

These lead to a third rule:

3. An ought is derived only from an ought, and an is is derived only from an is.


It follows from these two rules that the veracity of is and ought statements is determined very differently for each.

There are two very different forms of logic. The first - the logic of wants and desires - is called deontic logic. Other forms of logic (propositional logic, for example) describe the other sort of inference.

So why is all this important?

Well, as I said, before pretty much any enquiry, including scientific research, you need both an is and an ought. So, on the one hand, you need data, to tell you what is the case, and on the other hand, you need some sort of problem or domain of enquiry, which tells you why you need the data and what you hope to do with it.

This means that the citations for any research should include some from column A and some from column B. Good research requires a clear context, problem, or domain of application, and it requires facts, data and evidence. And - even more importantly - the references supporting each of these need to be of the right type.

4. Define context, problem or domain of application from expressions of need or obligation, including social, political and economic perspectives, and not from data.

This should be obvious, but isn't. Even your unit of measurement is going to incorporate these perspectives, and will in some sense define the desired state. The units of measurement are not inherent in the facts of the matter.

We often hear sentences like "the data dictates that...". No. The data does not dictate anything. The only thing a set of data can produce for you is more data.

For example, the data may say "5 percent of the people finish the course." Nothing about the quality of course design follows from this. You only get this sort of statement if you've already agreed that "not finishing reflects a design flaw" or some similar ought statement (which in turn needs to be substantiated).

I see in the academic community a lot of expressions of value or obligation criticized on the basis that it's not derived from data. The idea these critics express is that all reference in an academic paper ought to be peer reviewed, and the statements of value and obligation therefore grounded in some sort of fact. But that's an error. There is not some sort of fact-based mechanism for determining value or obligation. 

5. Define data in terms of empirical measurement, and not in terms of expression of need or obligation.

This is probably the most consistent flaw of research provided by the education policy 'think tanks'. The 'data' they provide owes as much to the center's political orientation as it does to 'facts'.

Take a statement like this: "The professional expectations for today’s teachers are undoubtedly high." This looks like data; it looks like a statement of fact (as indicated by the word 'undoubtedly'). But it's a statement of what ought to be, in two senses: first, it describes 'expectations', and second, it uses a relative value-laded term of measurement, specifically, 'high'.

Of course, it's OK to make statements like that. But they need to understood as expressions of what ought to be the case, and subject to assessment in terms of value and obligation, rather than represented as data and misused as the starting point for an action plan.

There's a lot more that could be said on the subject of ought and is, but I'll leave it here for now, happy if I've managed to alert the reader to be sensitive to these two types of statement.

More reading:

Is ought - University of Texas
The Is-Ought Problem - Wikipedia
The Is-Ought Gap - YouTube
Hume on Is and Ought - Philosophy Now
Is/Ought Fallacy - Fallacies Files


Share |
Views Today: 3 Total: 173.