The following text is from a chapter called “Concerning Knowledge,” from the book Philosophy of the Understanding by Andres Bello. The book can be purchased from Amazon. The book was originally written in Spanish and published in 1881. The text was translated into English by Alan. You can read the Spanish side-by-side with the English on Alan’s bilingual website. (image on the left: portrait of Andres Bello by Raymond Monvoisin; image on the right: portrait of Immanuel Kant)
Ideas, then, that do not come directly from observation, come indirectly from it through the faculties that we have enumerated, aided, if you will, by certain instincts that, to me, are revealed in the natural fluidity of the imagination.
But forming an idea is not the same thing as reasoning. Forming an idea of a centaur or a hippogriff is not the same thing as affirming its existence.
It is incontestable that, in all our knowledge, a lot of reasoning is involved, and the most important kind is that which nascent experience has not been able to give us, and that which is limited to the observation of individual events.
When I connect certain olfactory sensations to the same cause that produces in me certain tactile sensations; when I connect, for example, the flower that I hold in my hand to the fragrance that my nose perceives, I am using strictly experimental reasoning. But even propositions of this type (individual propositions, the first source of experience), assume certain instincts that, when generalized, later become principles or fundamental laws that preside over all intelligent acts. Such is the principle of causality, which connects every phenomenon to a cause; such is the principle of substantiality, which does not allow me to conceive a quality or a change without a substrate, without a substance in which it exists; such is the principle of non-contradiction, by virtue of which I cannot conceive that a thing is, and is not, at the same time; such is the principle of sufficient reason, by virtue of which I conceive that, with respect to everything that is, there is a reason for it to be what it is and not something else.
There is a particular constitutive principle of that experience, consisting of generalized observations. When I observe that when I rub wax with a piece of wool, it acquires the property of attracting the lightweight bodies that are nearby, and having done this experiment with the same result several times, I reason that in all of them the rubbing of the wax and the wool has produced and will produce the same effect, I take a great step outside the limits of observation, because it is evident that I could not think so if I did not proceed on the principle of the stability of the laws of nature; on the principle that I have called empirical, not because it is so, but rather because on it is based the generalization of all empirical results, the metamorphosis of an observed phenomenal connection into a constant phenomenal connection, into a necessary phenomenal connection. My observations undoubtedly enable me to know, as far as they go, the stability of certain connections; to extend this beyond the present sphere and even beyond the possible sphere of my observations, and even beyond the observations of the whole human race, and even beyond the observations of all finite intelligences; to reason that it is always established, and that it is impossible for it to fail, is a place to which neither perception, nor memory, nor imagination can lead us, without the aid of a tendency or spontaneous instinct to reason thus; nature suggests this reasoning; our reason accepts it, and we reason with the empirical principle, making it not so much a premise as an assumption of all our reasoning regarding phenomena. We can formulate it like this: Given the cause, the effect necessarily follows; that is, given the precursor phenomenon, the second phenomenon necessarily occurs.
“A thing cannot be and not be at the same time.” This is the formula of a fundamental proposition that is called the principle of non-contradiction; this principle is not only superior to the reach of observation, but absolutely necessary for all reasoning, for all rationalizations, for all knowledge. Suppose for a moment that such a principle did not exist: Nothing could be proven, nothing could be rationalized, nothing could be known. The law of non-contradiction is evidently a universal, irresistible fundamental virtue, which not only does not admit of proof, but does not allow anything to be proven or admitted without it.
The principle of causality is also necessary of absolute necessity. To begin to exist and to have no cause are ideas that contradict each other. That the principle of causality is derived from another (“Nothing produces nothing”), and is therefore not a true principle, that is, a supreme truth not derived from another; or that this second principle, as it seems to me, is nothing but the first, expressed in a different way, is a matter of little importance. We cannot but admit the absolute necessity of the principle of causality, whether we resolve it in another principle that we cannot fail to admit without incurring a contradiction, that is, without supposing that what is, is not, or what is not, is; or whether we find directly in it what compels us to accept it.
Another universal principle of absolute necessity is that of sufficient reason, which we can formulate in this way: “Nothing can be which does not have a reason for being.” This principle seems to coincide with that of the causal relationship, but with which, nevertheless, it is distinguished, as the logical antecedent is distinguished from the efficient cause. Suppose, for example, a line B, from whose extreme points, as centers, I trace two circles on a radius whose length is that of line C; the line joining the intersections of these circles evidently has each of its points at equal distance from the ends of line B, and divides it into two equal parts; because any contrary supposition would lack sufficient reason, that is, from logical antecedent, by virtue of the perfect symmetry of construction with respect to the ends of line B. The principle of sufficient reason is, in relation to logical dependence, what the principle of causality is with respect to phenomenal connections.
The principle of substantiality is also absolutely necessary. Given a change in a perceived phenomenon, it is impossible for us to stop assuming that there is a substrate, a subject, a substance, a modified thing. Having intuitively perceived in ourselves a substantial self, we make it an image, an idea-sign of all other substances.
There is a certain tendency of the understanding, superior also to all experience and necessary for the representative value of what we know by means of the senses. By the principle of causality, we connect our sensations to causes different from the substance that we feel. But in sensory perceptions, there is something more than this vague reference. Because of the relationships between sensations, we are inclined to represent relationships of the same kind between their causes. From the similarity of the sensations we infer the similarity of their causes. From the succession of sensations, and from their increase and decrease, we infer the succession and the increase and decrease of their causes. This, however, is a tendency, an instinct, rather than a principle. Thus, by blindly obeying it, we could often deceive ourselves; and in fact, not a few of our erroneous judgments concerning the phenomena of the material universe come from the fact that we yield to this tendency without examining it. Two similar sensations can sometimes be produced by non-similar causes: the same water at the same temperature will give us in certain circumstances the sensation of heat and in others the sensation of cold. The same thing that produces an intense sensation today can produce a weak sensation tomorrow. We have two successive sensations: A and B; their causes can be simultaneous; the cause of A may be perhaps not before but after the cause of B.
But how can we guard against this kind of erroneous reasoning, suggested by a tendency, by an evident instinct of understanding in the exercise of the senses? We cannot avoid it except by means of observations, that is, by means of the same senses, comparing and harmonizing the information from observation. Between the thunder and the lightning, we seem to have a more or less long interval. But the observation of a large number of facts reveals that sound propagates at a much slower speed than light and, having acquired this knowledge, we no longer reason that these phenomena are successive, even though the respective perceptions may suggest that we do so.
The special relationship perceived in the sensations supposes a foundation, a special relationship in the causes, but not precisely the same relationship: The deduction of this second relationship from the first is one of the goals of the physical sciences. They (referring to the previous example) tell us: The sensation of sound, insofar as it represents an aposcopic phenomenon, does not give us enough basis to believe that the aposcopic phenomenon is simultaneous with the sensation (simultaneously understanding that in which it is not possible for us to perceive an interval); the plesioscopic phenomenon, that is, the impression of the air in the organs, is what coexists with the sensation; and we are not authorized to deduce from the plesioscopic simultaneity the aposcopic simultaneity. On the contrary, there is always a greater or lesser interval, and sometimes perceptible, sometimes long, between the plesioscopic phenomenon and the aposcopic phenomenon; it follows that two sensations, whose causes are simultaneous, or appear to be simultaneous, may have aposcopic causes between which there is a perceptible interval and sometimes a long interval. We must, then, before yielding to this instinct, which literally translates, so to speak, the relationships of sensations into relationships of material causes, enhancing and reducing to their genuine meaning the report of the senses.
There are several things to distinguish here. Firstly, the relationships between the plesioscopic phenomena, although they have their basis in the aposcopic phenomena, do not involve in them relationships of the same type. Thus, the light that comes to us from a distant prism can make on our visual organs an impression similar to that of the light that comes from a cylinder. Secondly, the relationships between our organic impressions, although based on plesioscopic phenomena, do not necessarily imply relationships of the same kind in them: Different predispositions in the organs may give rise to different organic impressions, notwithstanding the resemblance of the external actions that affect them. Thirdly, the relationships between our sensations, although they have their foundation in organic impressions, do not precisely presuppose relationships of the same kind in them; for example, it is very possible that the different predispositions of the soul cause different sensations, notwithstanding the similarity of the organic impressions. When many causes converge to produce an effect, and when the causes alter or diversify the effect, it is necessary, so as not to deceive ourselves, to take them all into account.
By “experience” we mean not only the experience formed by the senses, but also that of the inner, spiritual world, in which the self contemplates itself; experience, by itself, that is, reduced to mere observation, has not been able to give us our first knowledge; our first knowledge has undoubtedly come with it; all knowledge chronologically prior to that nascent experience is a chimera. But, at the same time, it is incontestable that there is in the understanding a lot of reasoning and knowledge that is logically prior to the experience, which does not derive logically from it, neither by an immediate derivation, nor by a subsequent derivation, because there cannot be experience that does not involve them. “Every new phenomenon is a cause,” is a principle that could not logically come from experience. But, without an experience, it is impossible that this principle could have appeared in the mind. Let us distinguish, then, with Victor Cousin, the psychological antecedents and the logical antecedents of human knowledge.
We call empirical—or a posteriori—the reasoning and knowledge that logically derives from experience, assuming the principle of the stability of phenomenal connections; this principle, we repeat, is not in itself empirical, and we have only given it this name because it is implicitly contained in all our empirical knowledge. Suppose that this principle did not exist: Experience would not be of any use to us; it would be completely sterile; or, rather, what we call experience would not exist, for it is nothing more than knowledge of the general laws of the universe, deduced from observations by the empirical principle, which assumes the stability of phenomenal connections; the general laws of the universe would not be for us either general or laws, if we did not judge them as being stable.
We call reasoning and knowledge non-empirical, or a priori, when it is not derived logically from experience. Kant attributes these two characteristics to it: universality and necessity; these characteristics, as Cousin observes, are not identical. Necessity implies universality, but universality does not imply necessity. But if necessity assumes or implies universality, then we can omit one of these two characteristics: the one that is necessarily implied in the other. Non-empirical, or a priori, propositions, therefore, are propositions that we conceive of as necessary by absolute necessity.
The necessity characteristic of non-empirical propositions is an absolute necessity; in purely universal propositions, we conceive of a kind of necessity, but not absolute; it is not necessary of absolute necessity that sublunary bodies fall when unsupported; we can very well conceive of a different order of things. The necessity of empirical knowledge arises from the principle implied in it, from the stability of phenomenal connections. Assuming this principle, and the connection determined by its essential conditions, we consider it necessary. But its necessity is derived from that of the principle.
Is the principle of the stability of natural laws in itself a priori knowledge, in Kant’s sense? I think we should answer “yes.” But can’t we conceive that phenomena succeed each other fortuitously; that there are no stable connections between them? That would inevitably lead to the destruction of all empirical knowledge. The author of nature could have established diverse laws, of which we know some—thank heavens! But without the existence of certain laws—those that we know, or others—empirical knowledge would not exist. The certainty of the empirical principle is not due to experience; on the contrary, the certainty of experience is due to the principle that dominates and governs it, to the principle that, in this sense, we have called empirical, an abbreviated expression that signifies the principle of empirical connections.
But is it necessary of absolute necessity that phenomenal connections be subject to constant laws? Let us suppose that two phenomena are very precisely determined, in all their circumstances and conditions: the phenomenon A and the phenomenon B. Could it be that sometimes phenomenon A was followed by phenomenon B, and sometimes not? Isn’t there an obvious contradiction in this? It seems obvious that this is so. If we made that assumption, there would be no causes or effects; for the stability of phenomenal connections enters into the idea of cause, in whatever way the cause is conceived. Thus, the empirical principle, if it is not identical to the principle of causality, necessarily coexists with it.
That the Creator has subjected phenomenal connections to constant laws is an a priori principle and one of absolute necessity. But, first, the Creator was able to choose at will between these or other laws, and the choice that was made is a fact, or rather, a kind of fact to which we can only arrive at a posteriori; second, assuming a perfectly well-known a phenomenal connection, and supposing a perfectly well-known law established by the Creator at his discretion—the connection would not yet be necessary of absolute necessity: the Creator could suspend it, as he established it, acting, undoubtedly, in this suspension according to other rules of higher category, to which he desired from the outset that they were subordinated; there is nothing in this that is repealed by his infinite wisdom; nothing that should seem imperfect or inconsistent. Phenomenal connections, even if we were sure of knowing them perfectly, should not, therefore, be regarded as necessary, except in a highly limited sense, since, absolutely speaking, they could be interrupted. They are necessary in accordance with the general order, which is within the reach of limited human intelligence; they are contingent inasmuch as this order was flexible and arbitrary in the beginning, and insofar as this order is subordinate to another, vaster order, which we only obscurely glimpse, and which, from time to time, modifies or suspends the first.
In every proposition, there is a relationship, and in every relationship there are two terms. The idea of one of the terms can be so contained in the relationship that it is not possible to conceive of the term without the relationship. Reasoning of this type is called analytical because, in order to form it, it is enough to decompose the idea of one of the terms, between whose elements the relationship that is the object of the proposition is necessarily found. When it is said that bodies are extensive, we take from the idea of the body the idea of extension because we cannot conceive of an unextended body; if we analyze the idea that we have of bodies, we find in it, as a necessary element, the extraposition of its parts. And since analytical propositions do nothing but develop or explain an idea, Kant also calls them explanatory or illustrative propositions. They do not add anything to the idea, but they develop and explain it, and in this way they illustrate it.
Other times, the relationship is not necessarily included in one of the terms. “Unsupported sublunary bodies fall,” is reasoning of this type. It is true that it is a law of nature that unsupported sublunary bodies fall. But this law could, absolutely speaking, not exist: There is no contradiction between the idea of unsupported sublunary bodies and the idea of not falling. If an unsupported sublunary body does not fall, it does not necessarily mean that it is not an unsupported sublunary body. Relationships of this type add to the ideas of the terms something that was not in them; and from this addition comes a new idea, a new compound. The propositions that express it are therefore called augmentative and synthetic.
It is easy to see that analytical propositions rest on the principle of non-contradiction: To say that a body is inextensible would be the same as saying that a body is not a body. The truth of synthetic propositions does not rest on this principle. Does it always rest on the principle of empirical connections, as when I say that unsupported bodies fall? Or are there synthetic propositions whose truth is independent of experience? Let’s examine this idea further.
Cousin cites, as one example, this proposition: “Every change, every new phenomenon, every new existence supposes a cause.” If the reluctance of the understanding to conceive of a change without a cause is not resolved in the principle of contradiction, as it seems to me, in such a case it is necessary to admit that there are synthetic propositions in which the understanding adds to a term a universal and necessary relationship, independent of the principle of empirical connections. Kant calls synthetic non-empirical propositions synthetic a priori propositions.
Kant believes that all analytical propositions are non-empirical propositions, a priori propositions; and he believes, at the same time, that, of the synthetic propositions, those are empirical a posteriori whose truth rests merely on the principle of empirical connections; and the other empirical a priori, whose truth is not based on experience, like the judgment we just talked about, that there is no change without cause.
I believe that we can admit, without difficulty, that all analytic propositions are independent of experience, and that they only involve psychological experience; while every synthetic proposition is normally the child of experience, limited insofar as it is born of observations, and only general insofar as it has been extended by the principle of empirical connections. I say “generally” because I think it is necessary to omit the existence of synthetic propositions a priori, reduced, as I conceive, to the principle of causality, to that of sufficient reason, and to that of empirical connections. The rejection of any of these three propositions, even though the understanding cannot conceive it without apparent opposition, does not seem to me to involve contradiction.