Philosophy 152
Science & Reason
Spring 2006
Lecture Notes
Our next topic addresses a set of questions concerning scientific theories.
Preliminary point: set aside disparaging use of the word “theory”, as in “That’s just a theory” as said by a creationist about evolution. Theories can be good things, very reasonable to believe. Don’t think: because it’s a theory, it just as reasonable to accept it as to reject it.
I. The Issues
A. The Structure of Theories
Scientific theories are collections of claims used to explain, predict, organize, understand phenomena in some domain. Philosophers approach this, the way they approach some other things, by trying to come up with some systematic account of the way such collections of claims are structured or organized. How do the various parts fit together? This can raise difficult puzzles in logic that we will not address. But we will examine some of the central ideas. There is a view about theories that is the starting point for discussion, somewhat analogous to Hempel’s view concerning explanation. It is that theories are “axiomatic systems.” We will spell this out a little.
B. Reduction
There are a couple of related ideas here that we will discuss. For now, some examples will illustrate the topic. Suppose that we’re wondering about some political changes, e.g., a conservative gov’t comes into power. And you might give some explanation in terms of social trends and the general ways such things go. You might talk about unemployment rates, inflation rates, etc. and how they lead to changes in which party is in power. You might also talk about various groups - Catholics, or union workers, etc. But you might also think about it this: voter #1 voted for the conservative candidate for this reason, voter #2 for that reason, and so on. Here your claims and principles will be about individuals. And, you might think, once you’ve explained why each voter voted the way he or she did, you’ve said that those are all the voters, then you’re done. Some people have an inclination to say that the individual explanations are more fundamental than the sociological ones. In fact, you might get suspicious about the status of the groups and their properties. You might find yourself saying something like, “Really, there are just the individuals and their behaviors.” Somewhat the same questions arise if we think about the connection between the table and the littler things that compose it. Or gases and the molecules that make them up. All of this illustrates an inclination toward “reducing” things at one level to things at another. And this gets understood as reducing theories at one level to theories at another.
C. Theories and Observations
Another thing to think about - the one we’ll think about most - is this. Theories often invoke things that we cannot directly observe. They do not just report regularities that we observe. As a build-up to this, take a common sense example: prints in the snow in my backyard, I form the theory that the neighbor’s kid is cutting through. But I’ve never seen him. Seems reasonable. And I could get a camera to monitor. But now think about something like gravity. We notice things about the ways objects behave - falling to the ground, orbiting the sun, etc. Then we say that there is this gravitational force. But we haven’t seen it. So there is a question about the status of theoretical entities, especially given the idea that we are doing science which is supposed to so heavily tied to observation.
There are also questions about just what the theoretical/observational distinction amounts to. The neighbor’s kid was a “theoretical entity” at first. (At least for me. Probably not for his parents.) Some things in science are like this. But some seem not to be.
So there seems to be some kind of tension between two strands of our thinking about this. On the one hand, theories have to “go beyond” the observations to get at the underlying factors that account for our observations. On the other hand, it’s hard to see how any such step is justified.
II. The Traditional Account of Theories
A. Empirical and Theoretical Laws
There seems to be a distinction between the objects and properties that we can observe and those that our theories commit us to.
Observables: things and properties that can be directly observed. [Narrow interpretation: sensible qualities; broader interpretation: things that can be measured]. Let’s use the broader interpretation - things that can be measured in simple ways, e.g., temperature, pressure, weight.
It may be that what counts as observable changes over time.
Theoretical entities: things and properties that are not observed but are invoked in order to explain and formulate theories about observables.
There is also a distinction between:
Empirical laws: laws about observables
Theoretical laws: laws about non-observables (theoretical entities)
Examples: p. 74 - old theory about heat; p. 83 - phlogiston and combustion; p. 76 - kinetic theory
A theory (see p. 76) includes a bunch of laws, typically of both kinds. The theoretical laws imply that certain observations will be made in certain circumstances, and thus the proposed laws can be tested by seeing if the predicted observations actually occur.
B. Some Epistemological Questions
In thinking about theories, it will be helpful to separate epistemological and metaphysical questions. The latter are about what kinds of things exist and their properties and relations. The former are about our knowledge of the world.
Here’s an idea about how science proceeds: We arrive at empirical laws by inductive generalization. The idea is that a bunch of observations are made, a pattern is observed, and a general conclusion is drawn. You then come up with some theoretical generalization that accounts for this. The theoretical conclusion is not an inductive generalization. Three epistemological questions about this: Q1) How exactly does inductive generalization work? ; Q2) Why can’t inductive generalization be used to get to theoretical conclusions? Q3) How can we establish theoretical generalizations?
Q1) Inductive generalizations are generalizations based on specific observations. E.g.,
1. African violet plant #1 died after being left out in freezing temperatures for a month.
2. African violet plant #2 ...
....
_______
C. All African violet plants die after being left out in freezing temperatures for a month.
The key thing here is that the premise is entirely about observable properties - you can observe that the plant is an African violet, and that it died after being left out for a month. (Switch examples if anything about this one is troubling.) More cases in a greater variety of circumstances seems to strengthen the inference. There are lots of puzzles about induction. We will return to this later in the semester. Assume for now that it is ok.
Q2) But if inductive generalization involves generalization from specific cases of observed properties to generalizations about those same properties, then you can never get the terms of a theoretical generalization as a premise (or conclusion) of an inductive generalization.
But notice: the structure of the argument about theoretical properties could be the same. Suppose electrons are not observable. An inference such as: this electron weighs less than a 1 pound, ... to the conclusion that all electrons weigh less than a pound would have exactly the same structure as the others. It’s just that you can’t establish this premise by observation.
Q3) This is more complex. Here’s a description of what you might think happens. i) Upon examining some empirical generalizations, one comes up with a hypothesis, H. It’s not a logical derivation from the generalizations. Perhaps it’s just some sort of creative insight. (“Every combustible substance contains phlogiston.”)
ii) Then one sees that certain new empirical generalizations, EG, can be derived from H. (“Things decrease in weight during combustion.”)
iii) Then one tests to see if EG is true.
iv) If EG is true, then H is confirmed.
The reasoning involved in (iv) seems to be this:
1. H implies EG
2. EG is true.
3. H is true
v) But there are two complications: a) H does not exactly imply EG in the typical case. There are background assumptions playing a role (“Phlogiston has weight.”) There are other assumptions built in also, depending upon the details of the example.
b) And the argument is invalid!
So the pattern of argument must be more complex. We will return to this epistemological question later also. Let’s assume for now that somehow the reasoning here is good.