On Knowledge

My starting position is that knowledge is in heads, not on paper. I inderstand that I have personal 'internal' knowledge if I respond in similar ways to similar perceptions. The response may be unconscious - either sensory-motor or some internal conditioning of other states - or conscious, perhaps involving analysis or reasoning. At its simplest it is much the same as conditioned response.

Scientific knowledge
In these notes I am concerned with scientific knowledge. This could get recursive, since a generally acceptable definition of 'science' is 'a body of knowledge'. Although most dictionaries do not go into this detail, it is implicit that the body of scientific knowledge is shared by more than one person, and can be acquired by others. Which is why I equate scientific knowledge with transferable knowledge.

Transferable knowledge
In situations, which we agreed were similar, we would make similar judgements on causes and make similar predictions of consequences

Critical realism


So what is knowledge?


Explicit knowledge and tacit knowledge
I am primarily concerned (in these notes) with explicit knowledge. This is the sort of knowledge which can be told to others so that they can see the world in the same way, can repeat the experiments - the sort of scientific knowledge which can be written down and published. As a consequence, I talk of 'describing' the world. However, I suspect that much of what I say could apply equally to tacit knowledge, and even to sensory-motor knowledge (throwing a ball, riding a bicycle, ...) if one were to substitute 'interact, in a consistent way, with certain features of the world' for 'describing, in a consistent way, certain features of the world'.

Hard science and soft science

Transferable knowledge
Consider my question to Ph.D. candidates on 'making a contribution to knowledge': how would someone else know that you have made a contribution?

When I last looked at Popper's writings, I was not particular happy with his 'World 3' ideas, but I could live with them. It seemed to me that World 3 is there by cultural agreement. And one could decide whether to accept one set of World 3 knowledge or another. So, most people's World 3 is a mixture of that which they tacitly accept unthinkingly - as part of cultural conditioning - and that which they think about and 'decide' to accept from others, and maybe even from themselves.

So - that's OK. Making a contribution to knowledge means adding some models or model-fragments (or even a meme) to world three - i.e. finding some model which you - and someone else - find useful. (This is a 'pragmatic' position: there are others. See 'On epistemology')

By 'useful' I mean that others find your model (conceptual framework) has predictive power, and preferably predictive power in a form such that one might exert some control on the world so as to achieve some desired outcomes.

What does this entail?

Description and Prediction
First, the concepts in that conceptual framework must be recognisable so that someone else would (generally - not necessarily always) label features in the world, within the scope of your conceptual framework, the same as you would. [Note ...We usually think of 'a' someone else, but we must accept that a group effort, involving some social process, might be necessary to achieve consistent labelling. This is still valid, so when I use the phrase 'someone else' it means 'someone else acting alone, or someone or some people in a group performing some process'.]

Next, that someone else would then perceive the same relationships between the concepts.

Next, if we label some features of the world using the new concepts, then we - and others - will perceive that other features of the world also fit into the conceptual framework, mapped in the same way. And that is prediction. And the ability to make such a prediction is scientific knowledge.

[There are lesser forms of description: that which is entirely descriptive - that is to say that 'you' describe all the features of interest to your framework using the framework and then 'someone else' simply nods and accepts. Or 'you' describe some features of the world and someone else finds that description useful, appealing, satisfying, .... Such descriptions can be interesting, and when published they can even stimulate change in the world, but without any predictive power they are more in the realm of artistic criticism than scientific knowledge.]

More on prediction
The story so far ... You 'know' something when you have a conceptual framework and a way of using it to describe the world such that if you describe some perceived features of the world then you can predict that other features will be perceived. (I do not say 'pertain' because that implies an objective reality, and whether there is such a thing or not, I would rather avoid the presumption.)

Now much discussion of scientific knowledge makes use of examples drawn from dynamic modelling of the world in which time is an essential ingredient, and predicted properties of the world are necessarily in the future, after the initial description. Thus 'What goes up must come down (presumably later)'; 'The sun will rise tomorrow'. But what of Einstein's prediction of the bending of light by the Sun? This was only 'necessarily' in the future because we had to wait for an eclipse for an opportunity to perceive it.

It is not necessary for the futurity of a prediction to derive from temporal aspects of a model. You have been given the scientific knowledge that the external angles of any polygon sum to 360; you have an irregular pentagon in front of you; you have measured four of the angles; but although the remaining angle is already there, you have no option other than to measure it after the others.

The element of time associated with prediction is not, then, as important as is sometimes implied. Some predicted features of the world will be time-dependent, others not. Those patches of knowledge in which predictions are necessarily time-dependent because of (what we think we know of) the dynamic properties of the universe comprise a subset of all knowledge which seems not to be so special as we might have been led to believe.

But, taking that example of what we know about a polygon, when we say we 'know' that the external angles sum to 360, are we not just describing a property of polygons? Where is the prediction?

Knowledge as description that constrains
The prediction is implicit in the description. Having measured four of the external angles of your pentagon and found they sum to 320 you can predict from your knowledge that the remaining angle will be 40.

Knowledge is a set of constraints on our beliefs about what we might perceive of the world - in everyday language, a set of constraints on 'the way the world works'. (This gives real teeth to the saw "We are limited only by what we know.")

Knowledge is usually expressed in descriptions which embody, imply or entail the constraints, which in turn imply predictions. The descriptions are based upon some way of describing the world, and those descriptions generally embody some conceptual framework and some way of mapping perceptions of the world to that conceptual framework.

What of control?
If we want control then we have to do more than merely perceive features of the world from which we can predict that we will perceive others. We want to manipulate features of the world in accordance with such-and-such aspects of the conceptual framework, so that we - and other observers - will perceive such-and-such other features (or, in everyday shorthand, 'these other features will arise').

What do we mean by 'manipulate'? How about 'We interact with the rest of the world in such a way that, by luck or what we perceive as good management certain features are perceived'. Put these last two paragraphs together and you get 'If we interact with the rest of the world in accordance with such-and-such aspects of the conceptual framework so that certain features are perceived by ourselves - and other observers - then other features will be perceived'. What is the difference between these two sets of features? Not a lot, I submit. Within the constraints of the connectedness of reality and accessibility of 'control variables', there is no difference between 'input' and 'output' variables apart from whether they are chosen to be input or output in a given situation.

This is a consequence of the constraining nature of knowledge. We have simply moved from describing the constraints to using them to achieve 'control'.

Note that the 'and other observers' is critically important for such 'prediction' to count as knowledge, and it is, of course, the essence of the scientific 'repeatable experiment'.

GO TO .... | ideo home | SII home | EPSRC SI programme home | SI proposers' guidance