Making the weaker argument the stronger

Making the weaker argument the stronger

on
Anderson, J.R. (1978). Arguments concerning representations for mental imagery. Psychological Review, 85, 249–277.

Here I am taking issue with Anderson’s dismissal of the non-difference between neuro-physiological and behavioral data. If you are reluctant to hear that all data are theory-laden, I suggest a dose of Popper and Kuhn (Feyerabend if you’re feeling racy); and that you don’t read this as you probably won’t like it.

The hidden weakness of neural data

Anderson states that phsyiological data could present two kinds of worries to one wishing to use it to identify imagery vs. propositional representations of mental imagery. The strong one he argues is that nothing like direct evidence has yet been found for images or propositions. He adds

This is not to say that such data are impossible to achieve in principle, only that it is unlikely.

Anderson wrote this article in 1978. Right now is 2013_46_2_2127; a lot of work has been completed since then. According to Anderson’s claim, I will need to scour through the relevant literature to discover whether, indeed, such a demonstration has been found. I’ve not. I won’t. And as an argument I raise what Anderson describes as the weaker argument.

Namely, that non-serious argument is that ‘observations of brain functions have the same problem of interpretation that behavioral data do.’ His argument against himself is as follows

Suppose, not just to be bizarre, that we observed an m × n grid of data encoded on the brain’s surface and that this corresponded to a picture of an object. This observations is a datum that a theory must account for. It would be possible to attribute that datum to some source other than the fact that such acgrid was actually neurally encoded…such a level of skepticism is clearly unaceptable…direct observation always has had a priveleged status as a means of determining the state of an object, even though it must be handled with some caution (remember, eg..g, the stick that appears crooked in water).

What is particularly egregious here is that he almost builds the argument against himself, though he stops a bit short of laying it out extensively or (seemingly) with any genuine belief in his warning of caution. Were the phrenologists and mesmerists of the 19th century using the priveleged status of observation when they probed the heads of their ‘patients’ or waved sticks about a tree? Priestly had the data that convinced others that phlogiston was a substance not worth postulating. He observed the data that ‘disproved’ his theory directly; as directly as is possible given that he produced the data. And yet he still needed to interpret the data and that is where he faltered.

Similarly, it is impossible to say that an array of data could be “encoded on the brain’s surface” without saying something about what “encoded on the brain’s surface” would like like. But then to do so is a theoretical claim about the representations that underly neural structure (if not cognitive structure). That is, under some theories of neural activity (however bizarre), we have been given direct observations that there are images, or propositions, or both! Of course, you will need to tell me what it is that counts as neural activity, but I will just make the move that Anderson did and say that whatever theory you have for explaining your phenomenon is correct, except in that you’ve left out that whatever activity you have gathered is direct evidence for images, propositions or both (unless you’ve haven’t left it out, in which case I think you may have bigger concerns).

My point is not to suggest that there isn’t something called brain activity; or that nothing is “encoded” on the brain’s surface; or even that it is impossible that a m × n grid of data could be encoded on the surface of the brain. My point is that most of the important words in the last sentence are woefully underspecified without the “implicit ground rules” about what were acceptable variations in how information is represented neurally. Or, more accurately, my point is that your intuition about how the representation is to be read off from the brain is by no means obvious.

A final strand of this argument (one I admit I am reluctant to pluck) is that he allows for the doubting of introspective reports. In which case, I wonder how it is that he believes his direct observations are to be made. Did knowledge of the world leap past the phenomenological character of experience that he is dismissing as possibly untrustworthy to some deeper form of knowledge that cannot be so doubted? If he relies instead on more “objective” measures (like lines on a cathode ray tube) the question of how those measures were determined to be relevant of course arises. A theory shaped the creation of whichever particular measures are at our fingertips, and as valid as I’m sure they are, they then would be no better than the interferometer used to demonstrate (by Fizeau’s direct observations of light through an eyepiece after passing through water, lens and reflected off mirrors) that aether drags on light that goes through it.

The Fizeau experiment was valid and today is explained via relativistic velocities. His direct observations, in fact, weren’t taken even by him as direct evidence of the existence of this phenomenon, they required interpretation.

The success of the experiment seems to me to render the… hypothesis necessary, or at least the law…; for although that law being found true may be a very strong proof in favour of the hypothesis of which it is only a consequence, perhaps the [theory] may appear so extraordinary, and… so difficult, to admit, that other proofs … will be necessary before adopting it as an expression of the real facts of the case.

So for those neuroscientists hoping to end this debate on representation, I fear you are stuck with representations of your own, that render you unable to ever make the question obsolete. And whatever objections that are raised based on neural data laced with the traces of your favorite theory, I can only say… there will not and cannot be compelling reasons to accept your ground rules for the acceptable variations in these theories.