SARA H. VOLLMER
Assistant Research Professor
Department of Philosophy
University of Alabama at Birmingham
Birmingham, Alabama 35226
tel: (205) 934-4805
email: vollmer@uab.edu

"Two Kinds of Observation: Why van Fraassen Was Right to Make a Distinction, but Made the Wrong One"

Sara H. Vollmer

 

What distinguishes naked eye observation from indirect inference of objects on the basis of observation is that signals diffracted from the target are recombined by the eye lens to create images (showing shapes and orientations) of the object seen. It is this principle of diffraction from the object seen and recombination to display a representation of the object that gives epistemic priority to naked eye observation. I argue that this diffraction/recombination principle applies to many types of scientific observation, and that many objects that cannot be seen by the naked eye are as epistemically privileged as naked eye observation. More specifically, some entities that cannot be seen with the naked eye can nevertheless be observed on the basis of the same physical principle as those entities that can be. When any entity has been observed on the basis of this physical principle, I argue, it ought to have a similar epistemic status with any other entity observed in this way. This suggests that there is a principled epistemic distinction different from Van Fraassen's that might do the work van Fraassen intends his to do.

i.

In many, if not most, cases of scientific observation, when an entity is observed, no information about the shape and orientation of that entity is transmitted. When a gauge indicating the level of a particular solvent in an instrument is observed, for example, no information about the shape or orientation of the level of the solvent is transmitted. Or, when a chemist observes a chemical compound spectroscopically, e.g. observing caffeine through observing its infrared spectrum, information about neither the shape nor the orientation of the caffeine is ordinarily transmitted.

In contrast, in ordinary seeing, when an observer receives information about the entities he or she sees with the naked eye, the information ordinarily includes information about the shape and orientation of these entities; when we see an orchid with the naked eye, we receive information about the shape and orientation of the orchid.

When an electromagnetic wave is reflected by an object the wave is ordinarily perturbed by the object according to its inhomogeneities. That is, a field generated by inhomogeneities in the object causes a change or distortion in the energy distribution of a propagating wave front. The inhomogeneities in the object, then, are encoded by the propagating wave.

The propagating wave is, in the usual case, diffracted. This means that some part of the propagating wave is scattered, in a wave-like fashion, from each small region of the object. Any given propagating wave that is diffracted by a small region of the object is not propagated in just one direction: it is scattered in a wide range of directions, and so is spread out over a wide region of space. And, because the propagating waves diffracted from different small regions all scatter in a wide range of regions as a consequence of the diffraction process, the waves from each small region spread out so as to overlap with each other, each superimposing upon the waves from other small regions.

An image of the object can be formed. When an image is formed at the image plane, each wave that was originally diffracted in a wide range of directions from any small region of the object is no longer spread out over a wide region, and so is no longer superimposed upon waves spread out from other small regions. Instead, in the image plane, all the parts of a widely spread wave from one small region are superimposed at one small region in the image plane. Every part of the diffracted wave from a particular small region, then, arrives at one small region in the image plane.

So, each small region of the object correlates with some small region in the image plane. The spatial relation of each small region in the object (relative to other small regions of the object) is the same as the spatial relation of each small region in the image plane (relative to other small regions of the image plane). So, for example, all parts of the wave diffracted in a wide range of directions, say, from the tip of my middle finger are superimposed, in the image plane, on a small region corresponding to the wave scattered from the tip of my middle finger, i.e., its image. And that region is in the same relative location, i.e., between the image of my index finger and ring finger in the image plane, as is my actual middle finger with respect to my actual index and ring fingers.

The image is a particular kind of summation of a perturbed wave, a summation that displays in real space an inhomogeneity function of an object (or system), i.e., its structure. The perturbation of the wave so as to diffract it in a wide range of directions is called a Fourier transform of the object wave. The subsequent summation of the perturbed wave so as to form an image is referred to as an inverse Fourier transform of the perturbed wave. To form an image, then, is to determine, from the spatial distribution of a perturbed wave, the inhomogeneity of the object (Steward 1983; Strong 1958).

In ordinary visual observation, the inhomogeneity function is a function of the reflectivity of an object and/or the inhomogeneity in reflectivity between an object and its background. The propagating field, visible light, is perturbed according to the reflectivity of each small region of the object, so as to form an Fourier transform. The lens of the eye, then, refracts the light and forms the inverse Fourier transform of the object's perturbation of the light, that is, an image of the object.

Reflectivity is effective in creating an image because structures are inhomogeneous with respect to this property. That is, the structure of an object varies in reflectivity across the object, and/or the object contrasts with its background in reflectivity. Reflectivity is relatively high at the surfaces of objects (or systems), and more often than not varies across the object, being higher at some regions of a surface than others. Where light is absorbed by an object, reflectivity is low (relative to regions where light is reflected) and the corresponding region of an image is "dark". At spaces between objects, reflectivity is near zero.

When the perturbation of a wave occurs, as is the usual case, just at the surface of an object, what is encoded is a surface inhomogeneity function. Information about the interior of the object is, then, not transmitted, and so information about the interior of the object is not carried by the image. When an object is translucent, however, e.g., an ice cube, or a Lucite paperweight, the perturbation of a wave occurs not just at the surface of the object, but also in the interior. In this case, information is transmitted not only about the surface of the object at its exterior, but also about its interior, e.g., bubbles trapped in the ice, or figures embedded in the Lucite.

While the inhomogeneity function of a structure is, in ordinary visual observation, an expression of light reflectivity, this need not be the case. Objects are also inhomogeneous with respect to the reflectivity of other kinds of electromagnetic radiation (e.g., x-rays, microwaves, etc.), as well as being inhomogeneous with respect to density, or refractive index (the extent to which light is retarded on passing through the material). An image can, then, be an inhomogeneity function in reflectivity of electromagnetic radiation that is not in the visible range, or in another kind of intensive magnitude. In each case, the structure of the object with respect to the inhomogeneity is registered by the perturbation of a wave. Objects, then, can have more than one image, the formation of each of which, perhaps no more than any other, would count as determining the inhomogeneity function of the object.

So, the information encoded by the scattering of radiation can be recovered through the use of a lens, such as the natural lens formed by the eye. But this capability is not unique to the lens of the eye. It is also performed by lenses we grind, as in optical microscopes, and by specially constructed electric fields, as in electron microscopes. It, also, need not be performed through an actual recombination of the scattered radiation. The recombination function can, rather, be performed mathematically, as in x-ray crystallography, using an inverse Fourier transform.

ii.

In the Scientific Image (1980), van Fraassen explains that to accept a theory is to accept that the theory warrants certain beliefs in accordance with the theory. But, according to van Fraassen, opinions vary regarding which beliefs are warranted by the acceptance of a theory. Note that, to speak of "acceptance" of theories and the "beliefs" such acceptance warrant, means, first, justified theory acceptance. But, secondly, it does not always mean this literally. For what is almost always meant on van Fraassen's view is "qualified acceptance". Acceptance is sometimes qualified; it comes in degrees. Where "acceptance" is qualified, it warrants "beliefs" that are qualified, too (van Fraassen 1980, 8).

On van Fraassen's "constructivist empiricist" account of the kinds of beliefs a theory warrants, acceptance of a theory warrants the belief that what theories describe only about the "observable things and events in the world" are described correctly (p. 12; p. 4 referred to as "empirical adequacy"). On this account, then, the beliefs that can be justified by scientific theories about theoretical descriptions of entities differ according to whether the entity is an observable entity, or an entity that is not observable. (An alternative account van Fraassen calls the "scientific realist" account, wherein if an accepted theory describes a thing, then the expectation is that the description will be correct in all its details; acceptance of a theory on this account warrants a belief that all the things a theory describes, it describes correctly (p. 8).)

Before we can evaluate the epistemic distinction made on the constructive empiricist account, we need to understand what van Fraassen means by "observable". Certainly, what we see with the naked eye is observable. Van Fraassen says that seeing "with the unaided eye is a clear case of observation" (pp. 15-16). But are the things that we see with the naked eye the only things that are observables?

Van Fraassen notes that some things that can't be seen with the naked eye, such as the moons of Jupiter, can be seen "if you are close enough," and that these objects that could be observed, e.g., if you could get close enough, are observable. A "look through a telescope at the moons of Jupiter," he adds, "seems to me a clear case of observation, since astronauts will no doubt be able to see them as well from close up." (van Fraassen 1980, 16)

What, then, about objects that are too small to be seen with the naked eye? Hacking notes, in reference to van Fraassen, that "there is no way to see a blood platelet with the naked eye." We cannot, after all, "shrink to the size of a paramecium and look at it." So, on Hacking's reading of van Fraassen "we do not see through a microscope;" what is too small to be resolved with the unaided eye is not observable (Hacking 1985, 135). But there are other possible readings of van Fraassen. For example, van Fraassen says that observable is a vague predicate, and so in observation there is a continuum, with borderline cases. This leaves open the question not only of just which cases are the borderline ones, but also the question of the epistemological status of what can be seen under these borderline conditions. For van Fraassen, the borderline cases don't lie just at the intersection of naked eye and microscopic seeing. This is made clear by his suggestion that the electron microscope takes us even farther into the realm of unclear cases: we may find, he says, "a continuum in what is supposed to be observable: perhaps some things can only be detected with the aid of an optical microscope, at least; perhaps some require an electron microscope, and so on." (van Fraassen 1980, 16; my emphasis)

For the sake of argument, let us grant cases of optical microscopic and electron microscopic seeing as unclear cases, and consider entities that are farther yet along the continuum: ones that are too small to be seen with the electron microscope. This includes entities that are observable by x-ray crystallography; perhaps it is these objects, objects of a size resolvable by x-ray crystallography, that for van Fraassen mark the largest, definitively unobservable entities, among them a molecule of caffeine.

To take caffeine as a clear case of an unobservable would mean that, on van Fraassen's constructivist empiricist account of theories, no accepted theory that describes caffeine warrants a belief that what it describes, it describes correctly (for caffeine isn't observable). And, presumably, no accepted theory that doesn't describe caffeine warrants a belief about it. So no theory warrants a belief about a description of caffeine!

This would seem to be an unwanted consequence. Take, for example, our theories about the bond lengths and angles of a molecule of caffeine. While there are limits to the accuracy we can achieve, within these limits, we can probably describe the bond lengths and angles of caffeine correctly, e.g., by using x-ray crystallography. At least, there are a wide variety of reasons to think that this is the case, ranging from successful syntheses involving caffeine, to quantum mechanical calculations. An enormous body of scientific information attesting to the correctness of the structure - that is, the bond lengths and angles - gives us reason to think that we have got it right. There are about as many reasons to think that we have got the bond lengths and angles of caffeine right as there are to think that an artist-scientist gets a description of the orchid right when he or she produces a perspective drawing, or scale model of it. So, to make an epistemic distinction between them - that when the caffeine is described we are never warranted in believing what the theory describes; but when the orchid is described in a theory, say, about the growth rates of orchids, we are warranted in beliefs that what the theories describe, they describe correctly - would seem, if there is an epistemic distinction to be made, to be of the wrong kind.

To make this distinction by looking, first, to whether we can see something with the naked eye - which is where, insofar as we can tell, van Fraassen looks - wouldn't seem to be the right way to go about making an epistemic distinction. Of course, reference of some kind must be made to what can be observed by the observers in question: in van Fraassen's words, "what counts as an observable phenomenon is a function of what the epistemic community is (that observable is observable-to-us). " (van Fraassen 1980, 17, 19; his emphasis) Suppose, however, that my co-workers and I use a microscope in studying specimens. What is observable-to-us, then, could easily include, besides what is seen with the naked eye, what is seen through the microscope; why not? But this is not what van Fraassen means by observable. He claims - on the basis of what seems to be a natural language argument - that what is observable is what a group, e.g. humans, can observe with no instruments. The human organism, he says, "is, from the point of view of physics, a certain kind of measuring apparatus. As such it has certain inherent limitations," and it is "these limitations to which the 'able' in 'observable' refers - our limitations qua human beings" (p. 17; If this is the basis for his distinction, why, we might ask, do the borderline cases extend into the electron microscopic realm, when our limitations fall far short of this?). I argue that what is important insofar as epistemic distinctions are concerned is not our inherent limitations - not the ability to see just with the naked eye - but our ability to see with the assisted human eye. If the limitations of the naked eye are significant epistemically, an account of why this so needs to be given.

Sometimes it seems that van Fraassen grounds his epistemic distinction in the notion of "experience". He suggests, for example, that the scientific realist fails to understand the role that the limits of perception should play in determining our epistemic attitudes toward science, and so fails to understand van Fraassen's epistemic distinction. The scientific realist fails, van Fraassen suggests, because the notion that "experience is the sole legitimate source of information" about the world leaves him cold (van Fraassen 1985, 258). One consequence of this notion - what van Fraassen calls the "empiricist premise" - is that, since we cannot experience the future, experience is not a legitimate source of information about future events: in van Fraassen's words, it is not "possible to have a guarantee about the future on the basis of our experience so far", but only about what he calls the "actual" (van Fraassen 1985, 253). So, for example, when we observe a solvent gauge, what we experience is the solvent gauge; we don't experience the level of solvent in the tank. Were we to take the gauge as a legitimate source of information about the level of solvent, we would have to rely upon what we had experienced - e.g., the gauge having been right so far - as a guarantee about the future, something we cannot, as per the empiricist premise, rely upon.

But, accepting the empiricist's caveat that experience can't be a legitimate source of information about the future, it still isn't clear just how the notion of experience grounds van Fraassen's epistemic distinction between observables and unobservables. For instrument-aided observation can, and does, give experiential information. It can, for example, give precisely the same kind of experiential information about shape and orientation - e.g., the involutions, protrusions, and boundaries, say, of an amoeba or a mitochondrion - as ordinary observation gives about an orchid. This kind of information that we get through experiencing the shape and orientation of an object is information that is just about the object in the present; whether the information is a result of naked-eye or instrument-assisted observation, it can be highly detailed, and provide a rich visual experience. Interestingly, van Fraassen interprets the empiricist premise as applying only to observables: "experience can give us information only about what is both observable and actual" (van Fraassen 1985, 258; his emphasis). Consequently, for van Fraassen, experience is not a legitimate source of information about anything that is not a naked-eye observable. But why? Until we know why it can't be a legitimate source of information with respect to some kinds of instrument-assisted observation, experience hasn't yet grounded van Fraassen's epistemic distinction.

Van Fraassen's distinction, however, is suggestive. We wonder whether it isn't, in some way, right-minded. For one thing, van Fraassen privileges entities that can be seen with the optical microscope and the electron microscope. He does this, as mentioned above, by saying that these entities are at least borderline observables. But his reasons for this are unclear. While some things seen with the optical microscope are clearly borderline cases, - e.g., when whether we can see them depends on such things as how good our eyesight is, or upon whether we can manipulate them even though we might need a microscope to see them - not all are in this category (van Fraassen 1985, 254). And so much the worse for entities that can be seen only with the electron microscope, which are also inexplicably among the borderline observables!

However, van Fraassen's instincts - if it is fair to call it that - about entities that can be observed by optical and electron microscopy, I argue, are well-placed. As I have suggested, entities observed in these ways are of an epistemic kind with entities observed by ordinary observation. They all utilize the basic principle in which a wave is scattered, and that scattered wave is recombined in a particular way: through an inverse Fourier transform. As explained in part I, the inverse Fourier transform itself is in a sense no different when it is done by the lens of the eye, than when it is done in any of these other ways, recreating a display of the perturbation of the wave as caused by the inhomogeneities of the object. It is, largely, the method by which the inverse Fourier transform is performed, in addition to the size of the wave, that differs. So, whenever we observe by this principle of scattered waves, whether by optical microscope, electron microscope, telescope, or x-ray crystallography, we observe in a manner that is of a kind with ordinary visual observation. The entities we observe, then, might be expected to have a similar status epistemologically with every other entity we observe in the same way.

There are some kinds of objects that we can never observe in this way because there is a principled limit in size for this kind of observation. These things include electrons and subatomic particles. The smaller the object being observed, the smaller must be the wavelength of the radiation used in the observation. At some point, however, there is a problem. The smaller the wavelength, the higher the energy of the radiation. At a certain limit, the energy of radiation is large enough, and the object that is being observed is small enough, that the radiation, rather than being scattered, actually scatters - or displaces - the object, and the object can't be seen. This limit is reached at about the level of the individual electron. This alone is good reason to put entities such as individual electrons in a different class epistemically from entities, such as the atoms that collectively make up the caffeine molecule, which can be observed on the principle of ordinary visual observation. Interestingly, van Fraassen suggests that scientific realists do not fully appreciate that the models science gives us are unimaginably different from the world we live in experientially, and gives the example of atoms which, he says, have no "shape, place, or volume" except on hidden variable assumptions (van Fraassen 1985, 258). But the position of an atom can be taken as the position of its nucleus, and its approximate volume, and to some extent shape, can be determined from the distance and angles between the nucleus and other atoms. But electrons, on the other hand, do not have both a position and a momentum (or place, shape, and volume) on some interpretations, for these cannot be measured simultaneously. So, van Fraassen's suggestion - that whether an object has, as per our scientific models, shape, place, and volume, matters regarding whether it ought to count as observable - recommends the distinction I suggest in this paper, wherein groups of atoms can, but individual electrons cannot, be observed through the scattering of waves and the application of an inverse Fourier transform.

This permits us to draw a fundamental epistemic distinction. This new distinction, rather than being a distinction based on kinds of entities, or on the ways an entity can in principle be observed, is based on the way that it has been observed. Entities that have been observed on the basis of the same physical principle as ordinary observation are distinguished epistemically from those that have not yet been (or, as mentioned above, never will be) observed on the basis of this principle. So, for example, before an enzyme has been described using x-ray crystallography, it is ordinarily described using other, less detailed, spectroscopic techniques, operating on a variety of different physical principles. In such a case, exactly what is described, and so whether the description is a "correct" description of an object could, in a way, be judged to be less certain, or less well-grounded, than a description of an object on the basis of scattered radiation and an inverse Fourier transform. This is because when the physical principle on which something is observed is the same as that in ordinary observation, we can at least say that a description of an entity observed in this way is as correct as a description of an ordinary object, observed by ordinary means.

The reason for this confidence in the comparative correctness of descriptions of entities that are observed on the basis of the same physical principle is, in part, due to the fact that the description of an entity sometimes includes, or is affected by, a description of the physical principle. When the physical principle by which two entities are observed is the same, then the part of the description that the physical principle contributes is, at least, constant. This makes comparison of such observations, and comparative beliefs about descriptions of such entities, easier. An example is that if a description of the shape and orientation of an orchid can describe the orchid correctly, then a description of the shape and orientation of a molecule of caffeine can describe the caffeine correctly.

I have argued in this paper that the observation of any entity that utilizes the physical principle of the scattering of a wave, and the application of an inverse Fourier transform to form an image of the object, would seem should have the same epistemological status as the observation of any other entity made in this way. On this basis, then, there is no reason to call one entity, caffeine, an unobservable, and another, say, an orchid, an observable. Rather, if we want to say that an orchid is observable, then we ought to look to the principles - scattering and recollection of a wave - that make it visible to us in the way it is. Then, not only the orchid is observable, but also, the caffeine. Why privilege any one of these transforms over another? Or if we want to define observable in some other way, perhaps there is a distinction. But, in this case, we need to know the non-arbitrary basis of this distinction.

REFERENCES

Fraassen, Bas. C. van. (1985) Empiricism in the Philosophy of Science
in Paul M. Churchland and Clifford A. Hooks (eds.) Images of Science
Chicago: University of Chicago Press, pp. 245-308.

Fraassen, Bas. C. van. (1980) The Scientific Image
Oxford: Clarendon Press.

Hacking, Ian. (1985) Do We See Through A Microscope?
in Paul M. Churchland and Clifford A. Hooks (eds.) Images of Science
Chicago: University of Chicago Press, pp. 132-152.
Previously published, 1981, Pacific Philosophical Quarterly 62: 305-322.

Steward, E.G. (1983) Fourier Optics: An Introduction
Chichester: Ellis Horwood Limited.

Strong, John. (1958) Concepts of Classical Optics
San Francisco: W.H. Freeman and Company.