Hanson on observation

According to the conception of scientific method traditionally associated with Francis Bacon, science ought to begin with the accumulation of observations unbiased by any theoretical preconceptions.  The idea is that only such theory-neutral evidence could provide an objective basis on which to choose between theories.  It became a commonplace of twentieth-century philosophy of science that this ideal of theory-neutral observation is illusory, and that in reality all observation is inescapably theory-laden (to use the standard jargon).  That is to say, even to describe what it is we observe, we cannot avoid making use of theoretical assumptions about its nature, circumstances, and so on.

Norwood Russell Hanson’s 1958 book Patterns of Discovery is commonly cited as a classic expression of this view.  And one way Hanson repays study is that his treatment makes it clear that theory-ladenness does not by itself have relativistic implications, contrary to what one might suppose from simplistic accounts of contemporary philosophy of science.

To be sure, it can at first glance seem otherwise.  As Hanson emphasizes, there is a sense in which different observers operating with different background assumptions can be said to see different things.  One microbiologist looking through a microscope at what’s on a certain slide might see a cell organ, while another sees a bit of foreign matter that resulted from the staining process.  One seventeenth-century astronomer looking at the sun at dawn sees it move through the sky, while another sees the earth moving relative to the sun.  And so on.

Many would reply that in reality, they do see the same thing, but simply draw different inferences from it.  The assumption here is that there are two things going on, a visual experience on the one hand, and a separate act of interpretation of that experience on the other.  But this, as Hanson influentially argued, is an illusion.  One way this illusory assumption is sometimes spelled out is by taking two perceivers to have the same sort of retinal image, or the same physiological state, but then imposing different interpretations on it.  The problem here, as Hanson notes, is that a visual experience is simply not the same thing as the having of a retinal image or other physiological state:

Astronomers cannot be referring to these when they say they see the sun.  If they are hypnotized, drugged, drunk or distracted they may not see the sun, even though their retinas register its image in exactly the same way as usual.  Seeing is an experience.  A retinal reaction is only a physical state – a photochemical excitation.  Physiologists have not always appreciated the differences between experiences and physical states.  People, not their eyes, see.  Cameras, and eye-balls, are blind.  Attempts to locate within the organs of sight (or within the neurological reticulum behind the eyes) some nameable called ‘seeing’ may be dismissed… there is more to seeing than meets the eyeball. (pp. 6-7)

Another way the mistaken assumption in question is often spelled out is in terms of “sense-data.”  The idea here is that the two observers have similar sense-data – describable, say, as their being “both aware of a brilliant yellow-white disc in a blue expanse over a green one” (p. 7) – and that they then assign different interpretations to these sense-data.  But this is simply not a correct description of how visual experience goes in the normal case.  Consider the famous Necker cube example (Fig. 1 in the image above).  If shown such an image and asked what he sees, a perceiver might simply say “That’s an ice cube,” or “That’s an aquarium,” or “That’s a wire frame,” or any number of other things.  Context will likely play a large role in determining his answer.  For example, if the cube is part of a larger image that looks like a Scotch glass, he’s obviously more likely to describe it as an ice cube than as an aquarium.  The point though, is that there are clearly cases in which the way he’d describe his experience is as “seeing an ice cube” – as opposed, say, to “seeing a set of lines, and then going on to interpret it as an ice cube.” 

Of course, there can be cases in which the perceiver says the latter rather than the former.  In those cases there are two things going on, the experience and then a separate interpretation of it.  But there are not two things going in the ordinary case.  There’s just the one thing going on, seeing an ice cube.  Similarly, in the case of the two astronomers, one of them might describe his perceptual experience as “seeing the sun move” and the other as “seeing the earth move,” and with each of them there is just the one thing going on rather than an experience together with a separate act of interpreting it. 

Of course, the first astronomer would (as we now know) in this case be mistaken about what he sees.  But that is irrelevant to the point, which is that the content of his experience is correctly captured in the description “seeing the sun move,” rather than the description “seeing a yellow-white disc in a blue expanse” etc.  Here too we can certainly imagine eccentric cases in which what a perceiver sees could accurately be described in that odd way – for example, if the perceiver had no idea of what the sun is.  But obviously that is not what is going on with a normal adult, let alone an astronomer.  And if some astronomer did “react to his visual environment with purely sense-datum responses – as does the infant or the idiot – we would think him out of his mind.  We would think him not to be seeing what is around him” (p. 22).

Then there is the fact that, even when we are dealing with oddball cases like this, they still involve the application of concepts within the perceptual experience itself rather than in some separate act of interpretation.  We characterize the experience in terms of “a yellow-white disc” rather than “the sun,” but the characterization is internal to the experience rather than tacked on to it.  As Hanson writes:

The knowledge is there in the seeing and not an adjunct of it.  (The pattern of threads is there in the cloth and not tacked on to it by ancillary operations.)  We rarely catch ourselves tacking knowledge on to what meets the eye.  (p. 22)

Moreover, if perceptual experiences did not incorporate such knowledge, we would not be able to do with them the things we do in fact do, and which science and common sense alike require us to be able to do with them – namely, draw inferences from them and otherwise relate them to our larger body of background knowledge, including the scientific theories we test by way of perceptual experience.  To quote Hanson again:

Significance, relevance – these notions depend on what we already know.  Objects, events, pictures, are not intrinsically significant or relevant.  If seeing were just an optical-chemical process, then nothing we saw would ever be relevant to what we know, and nothing known could have significance for what we see.  Visual life would be unintelligible; intellectual life would lack a visual aspect.  Man would be a blind computer harnessed to a brainless photoplate. (p. 26)

This sort of observation is, of course, a longstanding part of modern critiques of naïve empiricism, from Kant’s dictum that “percepts without concepts are blind” to Wilfrid Sellars’ attack on “the myth of the given.”  And again, at first glance it might seem to entail a kind of relativism.  If what we perceive always reflects what we know (or take ourselves to know, anyway), but different perceivers have different bodies of knowledge (or of what they take to be knowledge), how can they ever perceive the same things?

But Hanson also makes it clear why such a relativistic conclusion does not follow.  To say that there is a sense in which different observers may see different things – as in the examples involving the object under the microscope, the sun at dawn, or the Necker cube – does not entail that there is not also a sense in which they see the same thing.  Suppose two people are looking at the cube and one says “That’s an ice cube!” whereas the other says “That’s an aquarium!”  It’s not as if there is nothing about which they agree.  We can imagine the conversation continuing: “Do you at least agree that there are lines there?”  “Of course!”  Here too what they take themselves to know is intrinsic to the experience rather than being tacked on in a separate act of interpretation.  But in this case it is shared knowledge. 

As Hanson writes, “if seeing different things involves having different knowledge and theories about x, then perhaps the sense in which they see the same thing involves their sharing knowledge and theories about x” (p. 18).  Disagreement takes place against a deeper level of agreement.  Indeed, unless there were some level of agreement (at least concerning what exactly the dispute is about) there couldn’t be a disagreement in the first place.  As Donald Davidson would later argue, even to make sense of what another person is saying as language (whatever he is saying in that language), we need to be able to tie his utterances to aspects of a shared environment, which entails shared beliefs about that environment.  Anticipating such arguments, Hanson notes that error is the exception rather than the rule: “We may be wrong, but not always – not even usually.  Besides, deceptions proceed in terms of what is normal, ordinary.  Because the world is not a cluster of conjurer’s tricks, conjurers can exist” (p. 21).

The theory-ladenness of perception thus by no means entails relativism or skepticism.  For it doesn’t entail that disagreement between observers need be total or even massive, or that all or most of our perceptual judgments might be in error.  Indeed (and to go beyond what Hanson himself says), it is consistent with holding that there is some level of theory, some very general way of carving up the basic structure of the world, that is common to all observers and which we cannot coherently deny to be a reflection of reality. 

Another key insight from Hanson is that the role language plays in the theory-ladenness of observation:

There is a ‘linguistic’ factor in seeing, although there is nothing linguistic about what forms in the eye, or in the mind’s eye.  Unless there were this linguistic element, nothing we ever observed could have relevance for our knowledge… For what is it for things to make sense other than for descriptions of them to be composed of meaningful sentences? (p. 25)

Part of this has to do with the general and essential connection in human beings between thought and language, a topic I examine in some detail at pp. 91-103 and 245-56 of my book Immortal Souls.  Our being rational animals – that is to say, animals capable of grasping concepts, putting them together into propositions, and reasoning logically from one proposition to another – goes hand-in-hand with our having a language with the semantic properties and combinatorial structure human languages have. 

Hanson’s point also has to do in part with the thesis that certain specific ways of conceptualizing the world are impossible unless certain specific forms of linguistic representation are first in place.  The examples he emphasizes have to do with how the formation of certain concepts in physics was possible only after certain mathematical techniques had been developed.  (I had occasion to discuss in a post from a few years ago the way the invention of novel systems of symbols makes possible new ways of conceptualizing things.)

In any event, it is crucial to emphasize that here too we are not talking about something entirely separable from and tacked on to perceptual experience.  Rather, having a linguistically expressible content is something inherent to distinctively human perceptual experience, at least in a mature and normal human being.  In human experience, conceptual and sensory content are fused, two aspects of one thing rather than an aggregate of a purely intellectual state (as in an angel) and a purely sentient state (as in a non-human animal).   

It may illuminate things to consider the Thomistic view that as a rational animal, a human being has a single soul that grounds both his intellectual and sensory powers, rather than distinct rational and sensory souls.  As Aquinas says in Disputed Questions on the Soul, “the sentient soul in man is nobler than that in other animals, because in man it is not only sentient but also rational” and again, “the sentient soul in man is not a non-rational soul but is at once a sentient and rational soul” (Article XI, at p. 147 of the Rowan translation).  If this were not so, then (the Thomist argues) a human being would not be a single substance, but a mash-up of two substances (as human beings are on Descartes’ account).

It is one organism, the human being as a whole, who sees something as an ice cube (thereby conceptualizing in the act of seeing) or sees that the earth is moving (thereby having an experience with a propositional content expressible in a sentence).  And it is in one act that these “seeings” are accomplished, rather than in two coordinated acts.  That is perfectly consistent with the obvious fact that we can distinguish between the conceptual and sensory aspects of the one act.  As Hanson writes:

‘Seeing as’ and ‘seeing that’ are not components of seeing, as rods and bearings are parts of motors: seeing is not composite.  Still, one can ask logical questions.  What must have occurred, for instance, for us to describe a man as having found a collar stud, or as having seen a bacillus?  Unless he had had a visual sensation and knew what a bacillus was (and looked like) we would not say that he had seen a bacillus, except in the sense in which an infant could see a bacillus.  ‘Seeing as’ and ‘seeing that’, then, are not psychological components of seeing.  They are logically distinguishable elements in seeing-talk, in our concept of seeing. (p. 21)

As the Thomist would argue, we can distinguish between a man’s animality and his rationality, but it doesn’t follow that his animality and rationality are grounded in two distinct substances only contingently united.  Similarly, we can distinguish between the sensory and conceptual content of a man’s seeing an ice cube or of his seeing that the earth moves.  But it doesn’t follow that there are two acts here, an act of seeing and a distinct, additional act of conceptualizing or interpreting what he is seeing.