Russell's epistemology Anti-Philosophy

‘Philosophy’ has never been Wisdom. Chapter 25: Bertrand Russell 3: What is Knowledge?

Thirdly on Russell, here is a summary of his Epistemology: his understanding of what knowledge is and how we acquire it.

It was first published in his book The Problems of Philosophy (1912), continued in his Our Knowledge of the External World (1914), and in his article The Relation of Sense-Data to Physics (1914).]

I am suddenly speechless on what to give as a punchy summary of BR on this subject. He didn’t say anything exceptional that one could disagree with.  Don’t we all agree that sensations come to us from the outside world, about what Knowledge and Belief and Justified Belief and Unjustified Belief mean?  He also went into the Definite Descriptions that Names are (see here).

Epistemology deals also with such questions as the difference between knowing how and knowing that; and between rational knowing and empirical knowing.  My head already droops at the verbal logicking in this subject.   

Anyway, here goes as far as Russell is concerned.  I have put his key concepts below in dense black.  He wrote that there are two ways of knowing objects — by ‘acquaintance‘ and by ‘description‘.  Acquaintance comes from ‘sense data‘, which are our immediate perception of colours, sounds, and so on. 

Everything else, including knowledge of the physical objects themselves, has to be reasoned to by the mind. (No, below he tells us that this kind of knowledge also comes by way of taking in what other people have said.)  He called this ‘knowledge by description'(from here).

That is the nub of what has remained influential from his epistemology, although he had later thoughts.  

To be acquainted with something is to be aware of it by sense-data.  When you see and sit on a red plastic chair, you become acquainted with its redness, smoothness, coolness and hardness.

And here is what Russell meant by ‘derivative knowledge’ or  ‘knowledge by description’: To know that this thing is called a “chair” and that it is often found with other “chairs” and with something called a “table”, requires us to make inferences from our general knowledge of facts and from our acquaintance with other similar objects.

Here is another example of knowledge by description: We know by description, and not by acquaintance with our sense-data, that Everest is the tallest mountain in the world.  We haven’t been there, let alone measured it and all the other mountains in the world.  So we have to rely on the testimony of others to “know” that Everest is the tallest mountain.

BR also gave the example of what we think we know about Julius Caesar (from here).  All of it comes by description.  When we use the term ‘Julius Caesar’, we don’t use it to refer to the man himself, but to descriptions of purported facts we have learned about him.  This also seems to foreshadow what Russell said under his later logical atomism, see here, that statements can be broken down into constituent assumptions. 

The argument is also tied to Russell’s Theory of Descriptions, see here, which explains how definite descriptions — such as ‘that cat’, ‘Bill Cosby’, or ‘my mother’ referring to specific objects—are shorthand for a series of logical sub-statements hidden inside the definite description.

The above is the nub of what Russell thought true Knowledge really is. 


All the paragraphs below, slightly paraphrased and edited to make them logically consistent in my own mind, are from sparknotes on Russell, perhaps all from here unless otherwise stated.  That latter website on Russell seems presently to have changed its subdivisions.  I have included these paragraphs for fear of missing out something important.  They appear at times in somewhat illogical and disconnected order.  It is the best my unphilosophical mind can do.  The whole lot is within square brackets. 

[In line with British empiricism, he thought that all knowledge comes from sensory perceptions, i.e. our sense-data..  These however can be erroneous:  For instance, if three people—one tipsy, one feverish, and one colour-blind—look at a table, they will see it differently.   If the object itself is under water or behind wavy glass, it will again be seen differently.  So how can we rely on sense-data?

But BR now deserted the strict logicking of Philosophy by accepting the commonsense view that all the viewers agree they are looking at the same table! 

Sense-data‘ was a term coined by Russell.  They are the mental images (visual, auditory, olfactory, tactile, and gustatory) we receive from the physical object.  Sense-data are related to the physical objects they represent, but the nature of this relationship is unclear. The skeptical argument through the ages was that sense-data tell us nothing about the reality of the object!

To the instinctive belief that the same table exists despite variable sense-data, Russell adds the hypothesis that the physical objects themselves actually cause the sense-data and therefore correspond to them in some way.  

The sense-data produced by physical objects are received and processed by our minds.  Russell calls the result ‘perceptual knowledge‘.  (‘Perception’, ‘appearance’, ‘experience’, all seem to mean the same.)

So Russell said that we have perceptual knowledge from sense-data mentally-processed  (i.e. a posteriori in that it comes after sensation) but Russell believed we also have a priori knowledge regardless of what comes to us from the outside world. These include the self-evident truths of logic and of mathematics.

Perceptual knowledge (of things out there in the world) and a priori knowledge (which he seems to have called ‘the knowledge of truths’) work together: the first gives us empirical data ( i.e. sense-data) and the second tells us how to process that data. (That last clause needs kindergarten explanation for me.)

Just as we know objects immediately by acquaintance or derivatively by description, so we also know ‘truths’ ( i.e. logical truths, i.e. a priori knowledge) immediately or derivatively.  Russell defines immediate knowledge of ‘truths’ as being intuitive. These are of concepts so clearly self-evident that we just know they must be true, such as ‘1 + 1 = 2’.  Derivative knowledge of truths however involves deduction and inference from self-evident truths.  (We need examples of these latter but I didn’t take any in from sparknotes on Russell.)

Some philosophers didn’t agree with BR that sense-data are the beginning of our recognition of objects.  They said that we start with immediate awareness of the table itself and only become aware of the sense-data if we then concentrate on them, (from here). 

Some critics also said that The Problems of Philosophy was an introduction, and Russell’s arguments weren’t thorough: that he often ‘illustrates’ his points rather than ‘meticulously mapping them out’.   

Also that there are unconvincing elements.  One such is that Russell never satisfactorily explains what exactly makes a truth (which presumably means a logical a priori one) ‘self-evident’ or  ‘intuitive’ or ‘immediate’ (all meaning the same thing) and that he doesn’t provide sufficient examples of them.  And that he provides no way of distinguishing between two apparently self-evident truths that contradict each other. (But I couldn’t find examples of these here.)


The following is a collection of BR’s thoughts as they appear on the Web, that I haven’t sufficiently understood, but they seem related to his epistemology, to his worrying about what Knowledge is and how we acquire it:


To start with (from here, and often verbatim):  BR wanted to be sure of the relationship between knowledge, and perception, and physics.  He accepted Descartes’ opinion that the foundation of knowledge is each person’s individual consciousness, Cogito ergo sum.  But then comes his criticism of the Cogito: how can a theory of knowledge for all people be built on someone’s private experiences?  Apparently for Russell, it was a difficult logical leap from the “private space” of personal sensation, to the “public space” of science and the physical world.  I don’t quite understand that but it’s probably obvious to philosophically-minded people.

Then comes this statement from here, which is perhaps relevant to trying to understand the previous paragraph: Sense-data are not simply images in the mind but are the building blocks of physics. Thus, sense-data inhabit the public space of science as well as the private space of experience.  (To my mind that seems merely to re-state the problem mentioned in the previous paragraph.)


Now I seem to return to something that BR has already said above (from here):

He had previously argued that ‘sense-data’ were caused by the objects and then perceived by our senses: for instance, a cat exists in the real world, and from it we sense warmth, softness, grayness. The problem is that we are then only acquainted with the sense-data—we have to infer that a cat is causing them, but we cannot know for sure that such a thing exists. (I would have thought that this theory, that there are only sense-data and that the existence of actual things has to be inferred from them, had already come up and been dealt with in the long history of Philosophy.)

Russell then argued that the most fundamental principle of scientific reasoning is: “Whenever possible, logical constructions are to be substituted for inferred entities.” This seems to  mean that we should stick to the logical reasoning that there is a cat and not assume that the cat does actually exist.  (I don’t understand this, but there it is.)

But BR now seems to do an about-turn or perhaps I am just dense:  Instead of continuing to say that physical objects create sense-data, he now argues that sense-data construct the physical object! Sense-data don’t just logically infer the existence of physical objects, but they essentially create them! (Is BR now saying that the sense data are all there actually is, and that there is nothing outside of them? This seems to conflict with what was said above, that BR was a realist who believed that the sensed world actually existed separately from our sensations of it.)


That latter new theory of BR’s goes along with what he now called ‘sensibilia‘.  These are ‘unsensed sense-data‘— how an object appears when no one is perceiving it.  This accounts for an object’s continued existence in the absence of perceivers. (Again that, to my unphilosophical mind, sounds like Idealism, even like Berkeley’s neat idea that God keeps an object in existence by always perceiving it.  BR had previously called Idealism  ‘poppycock’.)

Yet Russell was an Empiricist who at all other times believed that all knowledge must be acquired through the senses.  He also believed against the Idealists that the outside world existed regardless of whether we see it or think about it.  


The following two paragraphs are derived from here, often verbatim.

Russell shows how a logical atomism (which was his theory of how statements can be broken down into their unstated elements of meaning, see here) can also apply to knowledge of the physical world. This combines logical atomism with empiricism. The atoms of our knowledge are the sense-data of direct acquaintance.  These comprise the only knowledge we can claim.  ‘All other knowledge is inferred or deduced from it.’

‘Consider the cat.  Its sense-data—warmth, softness, grayness—are the “atoms” of our knowledge about it. The cat that we infer from the sense-data is only a “logical fiction.” Objects are nothing more than systems of sense-data.’  Hence there is  ‘Russell’s assertion that sense-data create rather than simply testify to the existence of physical objects.’!


And the next seven paragraphs are derived from here, often verbatim:

‘After advancing his theory that sense-data create the physical world, Russell abandoned it…and reverted to the notion that physical objects could legitimately be inferred from sensory experience.’   (So does BR abandon the idea that all we know are sense-data, and say now that an object actually exists independently of its sense-data? The quoted sentence is confusing as it seems to state both positions: that objects can be ‘inferred’ but also that they actually exist.) 

‘This [shift of BR’s] was partly due to advances in physics and physiology, which showed that perception is caused by the physical world on our sense organs.’  

Also, as I hesitantly understand, BR came to wonder (as we all do) about what he had called ‘unsensed data’:  How on Earth can there be sensation when no one is present to do the sensing?  And it also seems he couldn’t now understand what he had meant by  ‘private’ and ‘public’ spaces  or how sense-data and sensibilia interacted with them.  Oi!

‘In later work like The Analysis of the Mind (1921), Russell stops treating sense-data and the act of sensation as separate entities. He does, however, maintain the classic empiricist position that physical objects are not directly knowable; only their sensory effects… are… ‘     (Oi! say I, that’s Philosophy!  It seems too that philosophers can change their minds without notifying us that they’re doing so.)

‘Eventually, Russell abandoned his inquiries into the relationship between matter and perception, though he continued to work in other areas of epistemology.’

Other philosophers, including Wittgenstein, found flaws in BR’s idea that the elemental units of  knowledge of the physical world are sense-data, i.e. his logical atomism applied to epistemology.  (But this idea of BR’s seems to me so unexceptional that I don’t see how W. could have objected.) 

‘Although Russell’s….logical atomism may have proved untenable, it remains an important moment in the history of philosophy.’! (Again I don’t see how logical atomism, as a piece of linguistic analysis, is now untenable.) ]



Leave a Reply