by Robert Wicks
1 British empiricism: questioning the foundations of science
‘I’ll believe it when I see it.’ This familiar remark recalls how first-hand personal experience is commonly accepted as a convincing way – if not the fundamental way – to determine what to believe in. Many centuries ago, for instance, it was unbelievable that there could be organisms so small as to be invisible to the naked eye. Opinions changed after microscopes allowed us to see these tiny creatures in the late 1600s.
This emphasis upon direct personal experience is philosophically expandable to the point of prescribing how words must derive their meaning. The prescription would be: if any given word is to be meaningful, then it must be traceable back to some sensory experience. Otherwise, the word should be regarded as only a meaningless sound. This experience-tied theory of meaning was advocated by the Scottish philosopher, David Hume (1711–1776).
This does not mean that our words must refer only to things that exist. There may be no unicorns living in the world’s forests, but the word ‘unicorn’ is meaningful as the combination of ‘horse’ and ‘horn’ – items of which we do have direct experience. The name ‘Hercules’ is similarly meaningful as the enlargement of the idea of a physically strong person. The elements of our words must refer to things in actual experience, although the words themselves might not.
Hume was an empiricist philosopher, where an ‘empiricist’ is someone who holds that what exists or what is true in a non-trivial sense, can be known only through some observation about how the world is. To know anything, there must first be some sensory experience. The eighteenth-century British empiricists accordingly supposed that the mind is empty when experience begins, like a blank writing slate or ‘tabula rasa’, as the English philosopher, John Locke (1632–1704) described it. Sensory experience ‘writes’ upon our initially blank tablet, so to speak.
Since the empiricist outlook requires us to observe and experience the world before making any claims about what is true or what exists, one might expect it to be friendly to the scientific outlook, which also relies upon observation and experimentation. There is a surprisingly tense relationship, however, between empiricist philosophy and scientific thinking, for as we shall now see, strict empiricists have a doubtful attitude towards the concept of ‘causality’, upon which scientific thought is based.
At the foundation of scientific reasoning is the relationship of causality, for without this concept, it would be impossible to formulate any natural laws. We would like to say, for example, that heating water to a temperature of 100ºC under standard conditions will cause water to boil, or that the exposure to light will cause a chemical reaction on photographic film. Without a meaningful conception of causality, we cannot defensibly make such assertions.
Let us then consider how the concept of causality derives its meaning in empiricist terms – and most importantly, what kind of meaning it has on this empiricist view – by applying the definition of linguistic meaning mentioned above. Now science requires a concept of causality which expresses the thought that two events are necessarily linked. We cannot assert that water boils at 100º C and allow that under the exact same environmental conditions, it could boil one day at 100º, another day at 110º, and yet another day at 90º. If we assert that water boils at 100º C, we intend that water necessarily boils at that temperature, given how the world physically and constantly is. The claim about water’s boiling point predicts the future, which is what science is all about.
The scientific way, and also the common-sense way, to understand causality is to recognize that one event a is linked with another event b through a relationship of necessary connection. This gives us three components to the meaning of the word ‘causality’: (1) event a, (2) event b, and (3) a relationship between a and b – call it R – that is the ‘necessary connection’. We can write this in shorthand as aRb.
To understand the meaning of the word ‘causality’ in empiricist terms, we will need to experience each of these three elements individually and specify what they are. An easy test case will help, in which we can look carefully for each of these three elements in our experience. Imagine that we are watching someone bouncing a ball in a gymnasium. We see the ball hitting the floor repeatedly and we hear a ‘bouncing’ sound each time it hits. We naturally say that the sound is caused by the ball hitting the floor. Applying the empiricist theory of meaning, let us look for the individual experiences that correspond to the three components mentioned above, namely, the a, the b and the R. What do we observe?
The a and b – the ball striking the floor and the sound that follows – are perceived straightforwardly. The problem is with the R. The empiricist surprise is that aside from the a and the b, there is nothing objectively ‘out there’ on the gymnasium floor further to observe with respect to causality. We do not see the ball hitting the floor, and then observe some ‘link’ between the ball hitting the floor and the sound, as if there were a wire or chain connecting them together. Our direct experience of what is out there on the gymnasium floor, is of two events in succession, and only those two events. The R is not there.
Now there is an experience that corresponds to the R which completes the meaning of the word ‘causality’, but this is not the experience of an objective link that can be identified as ‘necessary connection’. It is something different. To find the R, we need to look inward and consider our own feelings, rather than observe what is happening out there on the gymnasium floor.
The experience associated with the R is the feeling of expectation that occurs when we see the a. In our example, it would be the expectation that we will soon hear a sound as we watch the ball moving towards the ground. Having seen many bouncing balls in the past, the expectation is a matter of custom or habit. It is psychological and ‘subjective’. Although it would be strange indeed, it objectively remains possible that the ball could hit the gymnasium floor and no sound would follow. Nothing precludes this. The future might not be like the past.
The unexpected result is that the meaning of ‘causality’ on this empiricist theory of linguistic meaning is not ‘necessary connection’, but ‘conventional association through custom or habit’. In our past experience, events have appeared in conjunction with one another, or have been ‘constantly conjoined’, but they need not have been, and they need not be so conjoined in the future.
This result goes a long way towards undermining scientific inquiry. The empiricist asks that we observe the world carefully in order to ground our knowledge, and observation shows that the concept of causality upon which natural science is based, lacks the strength to make solid predictions. The only legitimate meaning for ‘causality’, so it appears, is psychological association’ rather than ‘objective and necessary connection’.
Kant found this empiricist account of causality to be unbelievable, convinced as he was that scientific inquiry has a stronger basis than mere habit. Rather than accepting that empiricist philosophy had established that scientific theory is objectively groundless, he assumed that the problem resides within empiricism itself. Located among the theoretical possibilities that it could not recognize, was the correct way to ground the notion of causality as necessary connection.
Key idea: Causality as ‘necessary connection’
To ensure predictability between events, a scientific theory requires a concept of causality as ‘necessary connection’. Kant’s theory of knowledge intends to re-establish this concept by refuting David Hume’s sceptical conception of causality as the expression of simply habit and custom.
How, then, does Kant question and criticize British empiricist philosophy? His criticism is implicit in our initial chapter on Kant’s way of thinking. Specifically, he challenges the empiricists by casting doubt on their idea that all knowledge arises from sensory experience.
Kant asserts to the contrary that our minds are not originally like blank slates, blank pieces of paper or empty mirrors when sensory information begins to impress itself on our min
ds. Although our minds may be empty of sensory content before experience begins, they nonetheless have a prior structure that gives shape to the sensory experience. We are essentially rational beings according to Kant, and so logic and rationality are in us before experience begins. These give shape to our experience.
Kant develops his challenge to empiricism in a technical way by focusing on the empiricist theory of judgement, which we will now briefly describe. Since empiricists recognize sensory experience as the exclusive way to obtain knowledge of a non-trivial sort, they hold that our fundamental knowledge-carrying statements are simple matters of fact that refer to things in the world, their qualities or their relationships. Aside from this, there are only the definitions of words that we construct. These definitions are always ‘true’, but they are uninformatively so, since we construct the definitions ourselves. Matters of fact, when true, are substantially true because we have observed the world to be such.
In his An Enquiry Concerning Human Understanding (1748), Hume accordingly divides our knowledge-related judgements into two large classes: (1) relations of ideas and (2) matters of fact. Relations of ideas include geometry, algebra and arithmetic; matters of fact include observations that the sky is blue, the grass is green, and so on. These latter bits of knowledge are available only by taking a look to see how the world is. The philosopher, Leibniz, worked with a similar distinction between truths of reason and truths of fact.
Kant’s breakthrough was to ask whether the distinction between relations of ideas and matters of fact is fully comprehensive. Kant asked, because he was discerning that the distinction between relations of ideas and matters of fact was letting slip through the cracks, certain kinds of judgements that count equally as knowledge. Hume himself, who died five years before Kant published the Critique of Pure Reason, believed adamantly that his division was exhaustive, and he concluded his Enquiry with a dramatic remark to this effect. Any supposedly knowledgeable book that is not based on either relations of ideas or matters of fact, is suitable for the rubbish incinerator:
If we take in our hand any volume; of divinity or school metaphysics, for instance; let us ask, Does it contain any abstract reasoning concerning quantity or number? No. Does it contain any experimental reasoning concerning matter of fact and existence? No. Commit it then to the flames: for it can contain nothing but sophistry and illusion.
2 Kant’s theory of judgement: questioning the foundations of empiricism
In his effort to develop a theory of knowledge, Kant acknowledges the initial plausibility of Hume’s distinction between relations of ideas and matters of fact. Theorizing at a deeper level, however, he starts with the S is P format, since both relations of ideas and matters of fact assume this form. Kant’s preliminary concern is with the S is P logical format, where he asks us to consider the possible relationships between S and P.
There are only two options. In judgements of the form, S is P, either the meaning of the predicate P is contained in the meaning of the subject S or it is not contained in the meaning of the subject, and adds something new. Take the first option. If S is bachelor, then the concept male is contained in S’s meaning. The judgement, ‘the bachelor is male’, does little more than render a component of S’s meaning explicit. Judgements of this kind ‘analyse’ S, so to speak, and Kant refers to them as analytic judgements. They are true, but uninformative.
We should note that when saying that in analytic judgements, the predicate P is ‘contained in’ the subject S, it is easy to confuse what is meant by ‘contained in’. Consider this question: is the concept animal contained in the concept cat, or is the concept cat contained in the concept animal? One might at first think that cat is contained in animal, since animal includes many varieties, such as cats, dogs, sheep, goats, and so on. This, however, is not how Kant is thinking of ‘contained in’. To return to our example above, the idea would not be that male, contains bachelor, where the generic group of males contains both married and unmarried males, young and old males, and tall and short males. Kant has a different idea in mind.
The ‘containment’ situation compares closely to how a bowl of tomato soup always has both liquid and tomatoes. When we say ‘the tomato soup has liquid’ or the ‘tomato soup has tomatoes’, we render explicit the components of the concept tomato soup. We learn nothing new. Similarly, we could analyse the component parts of the concept triangle, and say, ‘the triangle has three sides’. In these examples, the idea of ‘containment in’ involves starting with a relatively rich concept – one that contains a number of components which together define the concept. The analytic judgements that express the contents of that richer concept, make explicit this or that aspect of the concept’s meaning.
Such is Kant’s understanding of analytic judgements, where the meaning of the predicate is contained in the meaning of the subject. Analytic judgements are consequently uninformative, since they only render explicitly what we already know implicitly. Aside from geometry, algebra and mathematics, which Kant will interpret differently, Hume’s class of relations of ideas corresponds to Kant’s class of analytic judgements.
We can turn now to the second option. This is when the meaning of the predicate is not contained in the meaning of the subject. Kant refers to these as ‘synthetic’ judgements, since the predicate adds new information that is not implicit in the subject. Unlike analytic judgements, all synthetic judgements are informative.
It seems that as of yet, Kant does not appear to have criticized or significantly modified Hume’s distinction between relations of ideas and matters of fact, since it appears that he has only expressed Hume’s distinction with a new terminology. It seems that Kant is simply using S is P as the basis of his account, along with the idea of predicates as either contained or not contained in the subject.
Spotlight: Questioning the analytic synthetic distinction
Up until the times of Leibniz, Hume and Kant, it had been taken for granted that there are judgements – Kant referred to them as analytic judgements – which, in effect, are true merely in virtue of the meanings of the words in the judgement. ‘A bachelor is an unmarried man’, for example, is true by definition. ‘Unmarried man’ is what ‘bachelor’ means.
Willard Van Ormand Quine (1908–2000), a philosopher working at Harvard University, questioned this idea in an influential, and often-cited essay entitled ‘Two Dogmas of Empiricism’, published in 1951. Quine carefully examined a series of attempts to define ‘analytic judgement’, including Kant’s, and found them all to be either vague, inconclusive, metaphorical or circular.
Since the time of that essay’s publication, it has remained a matter of vigorous debate whether there is a clear distinction between analytic and synthetic judgements. Quine’s view is that there is no genuine distinction, and that there are consequently no sharp boundaries between the disciplines of speculative philosophy and natural science. In later years, he stated accordingly that ‘philosophy of science is philosophy enough’.
The innovative aspect of Kant’s theory of judgement involves a second distinction, which when superimposed upon the analytic/synthetic distinction, yields some startling implications. The analytic/synthetic distinction essentially concerns whether or not a given judgement is informative. This second distinction concerns the different ways to determine whether or not a given judgement is true.
For any judgement, S is P, there will be items to which the judgement refers. Our new question is whether we need to examine those items to determine whether or not S is P is true. If S is P were ‘this bachelor is unmarried’, for example, and supposing the person referred to is a bachelor, it would be senseless to say to the person, ‘Excuse me, I heard that you are a bachelor, so I’d like ask whether you happen to be unmarried’. Once we know that a person is a bachelor, we know prior to meeting the person that he is unmarried.
If a man is a bachelor, we know a priori, as Kant would say, that the man is unmarried. Moreover, if a man is a bachelor, it is nec
essarily true that he is unmarried. Every such man is unmarried. So if we can know a judgement to be true without having to experience the items to which the judgement refers – if we can know a judgement to be true before experiencing the items to which the judgement refers – we know the judgement to be true a priori.
According to Kant, all a priori knowledge is necessary and universal. Every bachelor is necessarily unmarried and there are no exceptions. From the standpoint of philosophical propositions, it would be good to have some propositions in our theory that are known a priori, since these would be necessarily true and would apply universally.
In contrast, when we must experience the items to which the S is P judgement refers to know whether or not it is true, we are dealing with contingent matters of fact. Any one of them could be otherwise. If we were to assert, for instance, ‘This bachelor is 24 years old’, we would need to know something about the bachelor’s history to know whether the statement is true. Such judgements are known to be true (or false) only ‘after’ examining the items to which the judgement refers. Kant refers to knowledge gained by this kind of examination, knowledge a posteriori.
If we combine our two distinctions, (a) the analytic/synthetic distinction, and (b) the a priori/a posteriori distinction, some valuable results emerge that go far beyond Hume’s distinction between relations of ideas and matters of fact.
We can begin by drawing one obvious conclusion which concerns analytic judgements. If all analytic judgements are true simply by virtue of the meanings of their words, then to determine their truth, it is not necessary to examine the items to which the judgements refer. All analytic judgements are thus known a priori.
Suppose we defined a judgement where the subject already contains the predicate, but that to determine whether the judgement is true, one nonetheless had to examine the items in the world to which the judgement refers. This would be contradictory, for if the subject contains the predicate, we already know that the judgement is true. So there are no analytic judgements that are true a posteriori.