On Scientific Observation

Consider the following two statements describing the nature of scientific observation. (A) All observation is theory-laden.  And (B) all observation starts with raw-sense data.  I suspect that (A) describes that nature of scientific observation much better than (B), and will endeavor to briefly sketch out why.

In making any observation, we must make use of certain concepts that affect the way in which we see such-and-such.  For instance, if you draw someone a duck-rabbit and ask, ‘what do you see?’ they will either (1) see nothing (that is, apply no concept), (2) see a duck, or (3) see a rabbit.  But they cannot simultaneously see some combination of (1), (2), or (3).  So what you see is contingent upon what concept you choose to apply.  Here’s a more historical example. Aristotle saw a pendulum as a natural object that is ‘seeking’ to come to the natural state of rest.  He swung the pendulum, and gradually its period approached zero.  This confirms the theory that objects seek a natural state of rest.  Galileo, on the other hand, saw the pendulum as a natural object that was almost maintaining its inertia.  And the fact that a pendulum oscillates confirms that objects in motion actually ‘seek’ to stay in motion.  Galileo and Aristotle confirmed each of their contrary theories with the same exact data and observations because of the differences in their background knowledge and conceptual resources.

If observation got off the ground with sense-data then it is difficult to see how we could come to even see a duck or a rabbit in the duck-rabbit.  If we have no background theory, then it seems that we do not even have the resources to apply a concept to the image.  Without some concepts or assumptions about either ducks or rabbits, we can only ever see the duck-rabbit as pen-strokes on a page.  Ideas of induction and causation are not present in raw sense-data either.  But it seems that causation and induction are required in order to form a hypothesis like, ‘rocks fall when dropped.’  No finite number of observations in sense-data could justify this claim without presupposing a causal connection or the legitimacy of inductive inference.  It requires the assumption that something about rocks makes it such they they fall if unsupported.  These kinds of ideas must be part of the background when making some hypothesis — when articulating what it is that you are seeing.

There is an interpretive step between sense-data and any proposition you may use to characterize it.  Lakatos point out that propositions can be derived only from other propositions and cannot be derived from an empirical fact observed in sense-data.  There is a fundamental distinction between the data we strictly observe and the language that we use to articulate that observation.  Language immediately straightjackets our observations in the conceptual framework of thought or language — all empirical propositions are the recipients of some kind of interpretation (which will come from some kind of [primitive or not] background theory).  Consequently, any thinker or language user must see things in some interpreted way — for the language user, there is no uncorrupted sense-data, it is all theory-laden.

Moreover, if we hope to form any scientific hypotheses at all, then we cannot begin with raw observation, for we will have no way of selecting which observations to treat as relevant.  We must approach observation with some kind of expectation or background theory in mind in order to know what to look for, to know what kind of data we are interested in collecting.  This is a pragmatic point — we simply cannot note and observe a possibly infinite set of sense-data and then try to find regularities within that.  And even if we did, what sort of regularities we find depends on what sort of patterns we are looking for.  We must interpret data through some theory or other to even recognize which patterns in the data are regularities and which are just happenstance or coincidence.

Furthermore, science just isn’t about the raw sense-data.  As Schlick points out, science is about the discovery of laws and relations that exist between features of the given — it is not about the particular atomic features of the given itself.  Consequently, observation cannot start with raw sense-data, for we must actively impose laws and regularities of our own devising onto the atomic features or sense-data and observe what holds (as pointed out by Popper — we form a hypothesis and then seek to falsify it).  We are looking to observe what regularities hold in our experience/sense-data, we are not looking to observe a particular atomic feature of sense-data.  These regularities must be imposed by conjectures — and these conjectures provide the background theory that color our observation or commits it to certain concepts couched in our language (for all conjectures must be expressed linguistically).

So (A) is better.



One thought on “On Scientific Observation

  1. Pingback: The Priority of the System | Reflecting Light

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s