Some clarifications for a sentence from Bertrand Russell's work

Question

I just started reading Russell's "Human Knowledge: Its scope and limits." One of Russell's sentences struck me as particularly interesting. Russell said, I quote

...Inference from a group of events to other events can only be justified if the world has certain characteristics which are not logically necessary...

From what I understand, the phrase "logically necessary" means "a truth of absolute certainty". The term "justified" I understand it as "being deemed reasonable".

An inference from one event to another means the process of making a conclusion of some kind, right?

So putting all of them together, what is Russell trying to say, essentially? That we can't justify an inference if we know everything about the world with absolute certainty?

I'd appreciate some clarifications, thank you.

Edit: Upon further thought, I'm not sure what Russell meant by "a world having certain characteristics that are not logically necessary ..." either. Look, maybe it's just me... but why do I find that most philosophers' language are obscure and unclear?

Maybe I'm just too new to philosophy.

Answer 1

Russell is saying that there can be no purely logical inference from the existence of some events to the existence of others. You must assume, beyond logical rules, a law connecting these two groups of events.

To take an example inspired by your comments on another answer, let p be the proposition that there are some events (for example, that it is raining) and q be the proposition that there are other events (that the floor is wet). If you want to produce a valid argument (a justified inference) from p to q, you need an additional premise "p implies q" (if it's raining, then the floor is wet). This additional premise is what Russell calls "characteristics of the world which are not logically necessary". Indeed, it's not a matter of pure logic that when it's raining the floor is wet. You could imagine a possible world where the floor is never wet when it's raining for whatever reason, perhaps because the laws of physics are different there, and this possible world does not contradict logic (even though it is not our world).

As for the technical vocabulary, "necessary" is often analysed as "true in all possible worlds", so your intuitive grasp of logical necessity is correct, but more precisely, something is logically necessary if you cannot conceive a possible world that is logically consistent where that something is false, or, to say it more mundanely, a logical necessity "cannot be false, as a matter of pure logic". An example of logical necessity is the law of non contradiction: "either it is raining or it is not" (at a particular time and place). Laws of nature are not logical necessities, but physical necessities because we can conceive logically possible worlds where they do not hold. Hope this makes sense.

Answer 2

So, it's not entirely clear what Russell is up to in the quoted sentence, but here's a thought about what's going on.

I'll start with some background (if this is all familiar to you, even better). Russell is working with a philosophical picture of the mind according to which what is directly apprehended by the mind are 'experiences' or 'sense-data'. Knowledge of ordinary objects - tables, chairs, other people - is, on this picture, inferred from encounters with sense-data. Sense-data are known non-inferentially. So, for instance, my knowledge that there's a chair in front of me is inferred from the fact that I have brownish, chair-shaped experiences in my visual field, that the object resists my touch, and so on. The fact that I have brownish, chair-shaped experiences is known to me non-inferentially - it is just 'given' to me or is directly apprehended.

Now, given this picture of the mind it's very easy to generate skeptical worries. If all you're directly acquainted with are sense-data, then couldn't your sense-data stay the same while the objects on the other end of your sense-data differ? This is just Cartesian skepticism. If you're deceived by an evil demon, or if you're a brain in a vat, or whatever, then even if your sense-data represent there being a chair in front of you there really isn't - in reality you're dreaming, or you're in a scientist's lab, or whatever. Put another way, the worry is that your inferences from your experiences to objects in the world might fail.

But Russell tells us that he doesn't buy the skeptic's argument. He tells us that there are valid inferences

from events of which I am aware without inference to events of which I have no such awareness (HK, p. 2)

What does this mean? Well, Russell is telling us is that he thinks you can infer from your sense-data (these are the 'events of which I am aware without inference') to objects in the external world (these are the 'events of which I have no such [inferential] awareness'). That is, he thinks that the inference from 'brownish, chair-shaped visual experience' to 'there's a chair in front of me' is valid.

But the plausibility of this response hinges entirely on what Russell means by 'valid'. And I think understanding this is key to understanding the sentence that confuses you. One option is that by 'valid' he means 'logically valid' - that is, necessarily truth-preserving. An inference from P to Q is logically valid if Q is true whenever P is. So, on this reading, Russell thinks that the inference from 'brownish, chair-shaped visual experience' to 'there's a chair' is logically valid; as a matter of deductive logic alone, you couldn't have the experience without there being a chair too.

But this can't be what Russell means. First, he tells us as much:

if, then, I am ever to be able to infer events, I must accept principles of inference which lie outside deductive logic. (HK, p. 2)

That is, if we want to infer external events/objects from sense-data events, we need more than just pure logic. Second, he tells us that

as deductive logic can show, any collection of events might be the whole universe (HK, p. 2)

What he's saying here is that deductive logic alone doesn't let you discriminate between sense-data events and events in the external world. That is, you can't take your brownish, chair-shaped visual experience, somehow apply deductive logic to it, and get the conclusion that there's a chair in front of you (if there is) or that there's no chair in front of you (if there isn't and you're just deceived).

OK. So, deductive logic alone doesn't help us infer from sense-data to objects in the external world. Russell needs something else. He needs inferences that aren't logically valid, but which are still reliable enough to underwrite the inference from 'brownish, chair-shaped experience' to 'there's a chair in front of me'. So instead he's going to use inferences backed up by laws of nature:

All inference from events to events demands some kind of interconnection between different occurrences. Such interconnection is traditionally asserted in the principle of causality or natural law (HK, p. 2)

So, on this view, the inferences from 'brownish, chair-shaped experience' to 'there's a chair in front of me' is valid if there's some natural law or principle of causality that says 'if you have brownish, chair-shaped experiences (perhaps in normal circumstances), then you may infer that there's a chair in front of you'. This inference might fail (it's not logically valid, remember). But it's reliable enough to get us knowledge of the external world, which is what Russell is after.

Put another way, your inference from experience to external world is justified by there being a law of nature that says 'if you have brownish, chair-shaped experiences (perhaps in normal circumstances), then you may infer that there's a chair in front of you'. But this justification doesn't come from logic - hence your justification isn't a matter of the logically necessary features of the world. Rather, the justification comes from the non-logically necessary (i.e. contingent) features of the world, namely, the laws of nature that happen to hold in the world (remember: the laws of nature might have been different than they in fact are, so they aren't logically necessary).

Let's put all this together. When Russell says

Inference from a group of events to other events can only be justified if the world has certain characteristics which are not logically necessary

He means that our inferences from sense-data to events/objects in the external world can't be underwritten by logically necessary features of the world (namely, the laws of logic). Rather, they can only be underwritten by contingent features of the world, such as laws of nature. And obviously that can happen only if the world indeed has certain characteristics which are not logically necessary. 'Certain characteristics which are not logically necessary', in this context, means something like 'laws of nature'.

I hope that helps! If anything I've written is unclear please let me know.

Answer 3

Disclaimer: the best way to understand Russell is to find answers in Russell's own work because no one writes better English than Russell himself. The following is a clumsy attempt on my part. All praise goes to Russell; all errors, if there are any, are mine.

If P implies Q, then Q is said to be a logically necessary consequence of P.

Logic is the art of saying the same thing by different ways. If someone tells you to show up at 8:00am, and you ask "can I show up at 7:45am?", he will answer:" what did I say?" That is why Russell said:

What can you learn by means of deduction? Perhaps, if you are sufficiently clever, you could learn nothing. -- Russell. The Art of Philosophizing. 42. Print.

Inferences from events to events are not pure logical inferences. When you feel someone is pulling you up, you can infer that you are pulling him down with equal force. This inference implicitly assumes Newton's third law, which is a characteristic of the world, not logic. In other words, without the help of Newton's third law, logic alone will not tell whether or not there is an equal and opposite reaction.

Going back further, Newton's third law implicitly assumes the inductive principle, which lies outside deductive logic and is presumed to be a characteristic of the world.

Answer 4

Let's try an imagine a world in which everything is "logically necessary", and see whether inference from one event to another is justified.

It's a bit hard actually, but I think a good analogy is a movie or a book. Say you are a character in a book. In this book world of yours, a story is unfolding, events are taking place, and you are there to see them happen.

At one point in this story, you get shot. In the heart. The question you ponder immediately after is .... do I die now? Can I infer that I die?

In this book world, you actually cannot infer any such thing. Whatever happens is what was meant to happen to your character, what the author of the book intended and wrote down. Every event is logically connected via the words on the papers of a book. Does your character die? Live? Become an angel? You cannot infer anything: the only thing that will happen to your character is what has already been written down in the book.

So in such a world, inference from one event to another (the first event being "you get shot", the second event being "you die") is not justified. It makes no actual sense, and has no actual merit.