Sunday 18 May 2014

Inference & Sense-data: Davidson against Russell







Bertrand Russell (in his Problems of Philosophy) took a similar view to C.S. Peirce which smacks of the latter’s idea that ‘opinions will converge in the long run’[see Hookway, 1985]. Take the example of the ‘reality’ of a table:

"… the ‘real’ shape is not what we see; it is something inferred from what we see."

This is why the early Russell distinguished between ‘physical objects’ and ‘sense-data’:

"… if the physical sun had cased to exist within the last eight minutes, that would make no difference to the sense-data which we call ‘seeing the sun’. This affords a fresh illustration of the necessity of distinguishing between sense-data and physical objects."

There is a simple reason given here why sense-data and object are not, and can’t be, identical. Is this position that different to Berkeley’s point that "an idea can only be like another idea"? For a start, if we did not accept sense-data we should believe that a stick in the water really is actually bent. We may also think that the earth is flat (if we lived in Holland, that is). That the stars and the moon are much closer than they actually are. That train tracks converge in the distance. That the half moon is in fact, well, a half moon.

Even in these instances only, we do in fact infer from sense-data, Russell would have argued, to the reality of the object. But these are all examples of visual distortions. What about a plain and simple cognition of a table (as in Russell’s own example)? Do we infer from sense-data in this case?

Prima facie, it is certainly not obvious that we do. But surely sometimes we don’t cognise enough of an object to know that it is the object that we think it is. For example, when I see a table I do not see all four legs in one go, but I still know that I am looking at a table. Did Russell simply mean that we apply concepts and categories to what is, in fact, our insufficient sensory data? In that case, the inference to the existence of a table is carried out via the application of prior concepts to the table’s sense-data. These concepts, as it were, fill in the gaps left by insufficient sensory material. But what if we do have sufficient sensory data to work on? Perhaps then we would still need to apply concepts in order to come to a cognitive decision that it is in fact a table that we are looking at. Sense-data do not speak for themselves. They do not tell us what they are sense-data of. Of course we get to work on the sense-data; but they, in and of themselves, are not enough. Put that way, it seems obvious that sense-data cannot tell us what to think. As Davidson might have said: causal contact with the table is not enough to tell us that it is a table. We infer from the sense-data that we receive, care of concepts, etc., that it is, in fact, a table. Sense-data alone are not enough. After all, a native who is unfamiliar with tables may receive exactly the same sense-data as us, and have the same kind of causal contact with the table; but no amount of sensory material will tell him that it is a table in front of him. However, even if we do know about tables, we still infer from the sense-data, according to Russell, to the reality of the table. According to the half-full/half empty bottle example, it is our concepts and attitudes that help us decide what it is that we perceive. One person says the bottle is half empty. The other says that it is half full. Yet both persons are experiencing exactly the same sense-data. In this case, neither person is in the right or in the wrong. They are, in effect, letting their concepts and attitudes get in the way of a pure description of the bottle with water in it. Similarly, when we see the facade of a building - we automatically assume that it is a four-dimensional structure. Yet the sense-data themselves only ‘tell us’ that it is a facade – that is, something one-dimensional. We could, after all, be looking at the set for a western film in which the facades of the ‘buildings’ have nothing behind them. The well-known example in this context is the ‘elliptical penny’. The sense-data tell us that the penny is elliptical, whereas in fact it is round. It only appears to be elliptical, in this instance, because of the angle of our perception of the penny. If we were above the penny, looking directly down, we would indeed see a round rather than elliptical penny. Similarly, seen from a great distance, the penny would be neither round nor elliptical; it would be simply a dimensionless dot in the distance.

According to Davidson, even though we do not get sufficient sensory data, we still do not infer the existence of objects [Davidson, 1989]. We see objects. Indeed we will see tables even with a possible dearth of sensory data. This simply means that there are no inferential processes carried out by the percipient at all, even in the case of the elliptical penny. Concepts are applied instantaneously. There is no epistemic gap between the perception and the cognisance of the object in question. Davidson quite happily accepts the fact that there is a ‘causal gap’. That is, sensory data must come before the application of concepts. But the cognitive mind is not aware of this causal gap. In that sense, it is incorrect to use the word ‘data’ at all in that data is commonly perceived to be something that we work - or infer - from. It is the basis for cognitive work, as it were. If there is no cognitive work actually involved, then there is no data. Sense-data are in fact a myth. If anything we have object-data and event-data, not sense- data.

References and Further Reading

Davidson, D. – (1989/2000) ‘A Coherence Theory of Truth and Knowledge’, in Epistemology: An Anthology, eds. E. Sosa and J. Kim, Blackwell Publishers
- (1974/1984) ‘On the very idea of a conceptual scheme’, reprinted in his Inquiries into Truth and Interpretation, Oxford University Press
Hookway, C. – (1985) Peirce, London
Russell, B. – (1912) The Problems of Philosophy, Oxford University Press


No comments:

Post a Comment