Getting information: reason and observation
When knowledge is reasonable or sensible.
In previous essays, we have considered truth, and what makes something true (#10, #11, #13, #14); we have considered reality, and what makes something real (#15, #16, #17a, #17b, #17c, #17); we have considered personhood, and what makes something a person (#18, #19, #20, #21, #22). These are important questions, but there may be a difference between something being true, and me knowing that it is true, just as there may be a difference between something being real and me knowing it to be real; or between something being a person and me knowing they are a person.
Do we need to make this distinction? Surely, we could live in a world where my knowledge of the universe maps perfectly onto what the universe is like: I know I had rice for dinner, and I really did have rice for dinner; the world could be identical to my knowledge of the world. Before we rush to embrace this possibility, we must notice some difficulties it raises: I have no knowledge of what you had for dinner. If reality matches my knowledge perfectly, and if I have no knowledge regarding what you had for dinner, then there is no reality regarding what you had for dinner. Maybe that really is how the universe works. Maybe not. But there is certainly a case for carefully thinking through how knowledge (epistemology) and being (ontology) relate.
To this end, there are three questions we shall consider:
- How do we get information about the universe? (Here and #23b.)
- How do we put the bits of information together to draw larger conclusions? (#24.)
- Do we have any reason to believe our conclusions are correct? (#25.)
It would be nice if there was one single way that we got information about the universe. You know, something like “I know something happened if and only if I saw it happen.” But there is not one single way.
We have multiple, independent, irreducibly different ways of obtaining information about the universe. There are three broad categories that we shall discuss here:
Setting reason and observation as basic ways of gaining information and knowledge of the world might seem unsurprising enough. Setting faith on this list may seem a little more surprising. We shall therefore first look at how and why reason and observation managed (historically in the West) to muscle faith out of the picture. We shall then consider how reason and observation relate to each other, and how they relate to science and to religion. In Essay #23b we will unpack how this framework permits us to se that faith — far from being a cop-out for people who do not have information — can be seen as a way of obtaining information about the world.
Reason and observation alone
Our modern surprise that faith would turn up on this list can be traced back to the start of the Enlightenment. Consider the words of David Hume:
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. 
Hume here sets up two categories: abstract reasoning and experimental reasoning. These come down to us today as reason and observation; theory and experiment. Reason provides us with necessary truths (like “triangles have three sides”). Observation provides us with contingent truths (like “I am in Hong Kong”). Having so neatly carved up human knowledge, Hume seems to have all the bases covered. If reason and observation cover all the bases, there is nothing that can provide knowledge which cannot be understood in terms of these two. Thus, any statement in any book of divinity is either traceable to reason, or to observation, or to some combination of the two, or else the author has no way to claim they know what they are writing. If the author did not know it, why did they write it? Into the flames it must go.
This creates two sets of dichotomies for knowledge: reasoned, theoretical, a priori, necessary knowledge on the one hand; and observed, experimental, post priori, contingent knowledge on the other. If you have been following these essays from the start, you may expect that a neat set of dichotomous relationships like this is just asking for us to object to it. And we shall. Reason and observation have a tremendously interesting and complex relationship.
Hume’s fork also feeds into the dichotomies that we encountered back in Essay #1: reason and observation give us knowledge: objective facts, reliable and testable, worth having. Anything else does not give us knowledge, but instead subjective beliefs, unreliable and untestable, not worth having. Again, you may expect us to object to such. And we shall. Reason and observation need not be the only ways to apprehend things which are worth knowing, and they may even be less reliable than the alternatives.
With that set up, let us consider reason and observation in more detail.
Knowing by reason and observation
Reason is where you come to conclusions by the mind: use of logic, maths, rationality, reflection, and the like. Observation is where you come to conclusions by the senses: use of sight, hearing, proprioception, kinaesthesia, and the like.
Reason and observation are distinct. Reason is not some special kind of observation. It is not watered-down observation. It need not even rely on observation. It stands on its own. Likewise, observation, is not some special kind of reason. It stands on its own. As reason and observation give two separate ways of knowing, let us ask, first, which way is better?
Theorists like reason. Experimentalists like observation. Rationalists believe that reason is superior: conclusions attested by sound logic are surely unassailable. Empiricists believe that observation is superior: experiments really can refute theory. Despite the partisan lines being drawn, both reason and observation have their place.
Some statements can be demonstrated by either observation or reason. A child may initially learn mathematics by observation: They take one Lego block and one more Lego block and, putting them together, they find they have two Lego blocks. They have experimentally demonstrated that (as far as it applies to Lego blocks) 1+1=2. Professional mathematicians can pull out set-theoretical proofs to demonstrate that one plus one does indeed equal two, without any recourse to Lego. Both observation and reason can get you to the same conclusion.
There are some statements that could, in principle, be demonstrated by either reason or observation but which, in practice, are only ever demonstrated in one way or the other. Consider the claim that, if there were ten billion people in the world and one more person was born, there would be ten billion and one people in the world. There is no fundamental reason why you couldn’t demonstrate this experimentally: Have a campaign to increase the world’s population to ten billion, then count the population after one more person is born. Such an empirical effort would, however, be a lot of work. No one would begrudge you taking the reasoned short-cut of pulling out a pencil and paper and asserting — even if you had not observed it directly — that one billion people plus one person is one billion and one people.
There are some statements which can be justified by reason, and which can only be justified by reason. Consider the claim that there are infinitely many positive integers: 1, 2, 3, 4… No one has ever experimentally demonstrated that, if you kept counting for eternity, you would never run out of numbers. No one could ever demonstrate it experimentally. Nonetheless, the infinite nature of the integers is simple enough to demonstrate by reason, and that demonstration is considered acceptable.
There are some statements which can be justified by observation, and which can only be justified by observation. Consider the claim that putting your finger in a flame is painful. All the maths in the world cannot show this to be true. Nonetheless, the briefest experiment of putting your finger in a flame will demonstrate it to be so, and that demonstration is considered acceptable.
There are some statements which cannot be justified by either reason or observation on their own, but which can be justified by reason and observation together. If a scientist wants to know the resistance of a piece of wire, they must invoke both theory and experiment. A theoretical calculation alone, without any observational input, can never tell me the resistance of a particular wire. Without — at very least — knowing (observationally) that the wire in question is 10 cm long and 1 mm in diameter, theory has nowhere to start. Equally, an experimental observation alone, without theoretical considerations, can never tell me the wire’s resistance. I can see the wire’s length; I can feel its weight; but I cannot directly sense its resistance without some intermediate theoretical steps. It is often difficult to separate out how much of a measurement is, in fact, sensory observation, and how much of it is rational reasoning. This is generally unproblematic, because the exact divide is usually unimportant.
Both reason and observation seem to provide us with access to information about the universe. Still, we might still ask which is better? What should we do, for example, when reason and observation seem to disagree? Which way of knowing takes precedence?
If a scientist measured a room-temperature wire to have zero resistance, then theoretical considerations dictate that this is almost certainly impossible. It would seem more likely that the apparatus is faulty: reason trumps observation. If, however, the observation is corroborated using different methods, it may be accepted that the observed result is correct (and Nobel-Prize-worthy) even without theoretical explanation: observation trumps reason.
It is in no way considered problematic that there exist two entirely separate and yet equally valid ways of obtaining knowledge. These ways of knowing are irreducible: reason is not a particular type of observation, nor is it based on observation. Similarly, observation is not a particular type of reason, nor is it based on reason. They are different. Nonetheless, either separately or in conjunction, they are considered to constitute evidence. It is not considered problematic that these two different ways of knowing do not always point to the same conclusion. It is also not a problem that, when they disagree, we cannot always immediately tell which one is right. Science does not collapse. We have theorists and experimentalists, and science goes on.
Reason and observation get more complicated
This picture of reason and observation providing two distinct yet intertwined ways of obtaining knowledge is, of course, too simplistic. If someone dropped a book on my desk, I can know this by observation: I saw it happen. If, however, I were blindfolded so I could not see it, I might still hear it land. Moreover, depending on the book, even if my ears were plugged I might still have felt the desk shudder as it landed. Seeing, hearing, and feeling are all manners of observation — none is reducible to reason. But, more than that, one kind of observation is not reducible to another kind of observation. Feeling is not reducible to hearing, and hearing is not a form of seeing.
Some things, unlike falling books, are knowable only through one sense or another: the squeak of a door cannot be seen or smelled. Sometimes different senses can work together: it is easier to understand what a person is saying when you can both hear them and see their lips, compared to when you must rely on sound or lip-reading alone.
Reasoning is just as varied in character as observation. One can reason numerically, geometrically, or temporally. Just as a person may be deaf but perfectly able to see, a person with acalculia may be unable to reason numerically, while being perfectly proficient in geometrical reasoning.
Sense information and reason can work with each other to provide information normally limited to another sense or other modes of reasoning: if you see someone screw up their face, and you empathize that their responses are like yours, then you can infer that their food tastes sour, without ever tasting their food yourself.
At other times, senses can work against each other, or against reason: a surface may be coloured to look rough, while it feels smooth; an optical illusion may seem to present a shape or an action which we know by reason to be impossible.
It might have been nice if we had a single, unique, unassailable method of gaining information about the universe: some gold-standard benchmark which suffered no competition and against which there was no appeal. In that kind of world, any other “ways of knowing” which attempted to usurp the total supremacy of that one ultimate method might be immediately rejected as pretenders to knowledge. But we do not live in that world. We live in a world where reason and observation happily co‑exist side by side, as do multiple manners of reasoning and multiple manners of observing.
Reason and observation get weird
As we investigate reason and observation, we find that the further we look, the more varied these ways of knowing are. Touch is not one undifferentiated sense: the human body has separate sensors for feeling (among other things) texture, pain, and heat. Even then, feeling heat is not one thing, but the body has separate sensors for cold, warm, hot, and painful. Then there are numerous other senses quite distinct from what we consider under the usual five. Interoception tells you that you are hot (which is quite distinct from knowing that your surroundings are hot). Proprioception tells you that your elbow is bent. Kinaesthesia tells you that your elbow is bending.
While the senses considered above are generally possessed by everyone, not all people posses the same senses.
Some people have fewer senses than others: People who are deaf cannot hear things that other people can. People who are colour-blind cannot distinguish all the colours that other people can. (Typically this is because, while most people have three types of photo-receptive cells in their eyes, colour-blind people only have two.)
Some people have more senses than others: tetrochromats have four types of colour sensitive photo-receptors in their eyes, rather than the usual three. They are to ‘normal’ people as ‘normal’ people are to colour-blind people. (Women are much more likely to be tetrochromats than men. It may be that the reason your girlfriend agonises over colour swatches labeled Chalk, Alabaster, and Chantilly Lace, while you know they are all just beige, is that she has a super-power that you lack.)
Some people connect senses and reason in ways that others do not. Possibly one of the strangest of these is so-called blind-sight. People with this condition have functioning eyes, but have no experience of seeing because, for what ever reason, their brain is not able to process the optical signals in a normal way. Their brain instead routes the optical signals to their subconscious, permitting them to act on visual cues of which they are consciously unaware. Consequently, if they are asked what letter is on a flash card in front of them, they will reply that they do not know (because they have no experience of seeing it). If you ask them to take a guess anyway, they will “guess” correctly (because their brain can see the letter, even if it has not told their conscious experience about it).
Looking beyond humanity, we also know that the senses we have do not exhaust the limit of the senses that can be had, or even the senses that we could have. Some snakes can see in infra-red. Some insects can see in ultraviolet. Mantis shrimps can see polarisation. Bats can see (hear?) with ultra-sound. Sharks can feel (?) electric fields. The fact that humans don’t have shark-like ampullae that would let us detect electric fields is not a necessary fact of biology, but a contingent quirk of how things happen to be.
Looking beyond life itself, computers pose interesting questions for how non-human reason can contribute to human knowledge. It is clear that computers are able to use reason. It is also clear that computers are to perform more calculations than a human can, simply because humans are slow and prone to die before completing long calculations. What is less clear though, is what status should be given to a reasoned calculation performed by a computer which a human is not able to repeat.
The problems caused by advances in computer reasoning were sharply illustrated by efforts to prove the four-colour map theorem. The theorem states that, using only four colours, any map can be coloured in so that no two adjacent countries are the same colour. After many attempts to prove it, a proof was offered in 1976 in the form of a computer program. A human is able to understand what the program is doing, but no human is able to perform the required calculations in a single lifetime. When the program is run, the computer performs the calculation and returns the answer, “Yes. The theorem is correct.” As such, the four-colour map theorem is known by reason, but not directly by any human reason.
Both reason and observation have a place in science. That conclusion in itself is deeply unsurprising. The interesting things come in the details regarding what we mean by this.
The fact that reason and observation cannot be reduced to a single path to information means that science must be open to a plurality of ways of knowing.
The fact that neither reason nor observation infallibly provide us with knowledge of the world means that neither provides a final arbiter for scientific evidence.
The fact that not all people share the same faculties of reason, and not all people share the same faculties of observation, means that we cannot hope for universal standards of demonstration.
The fact that our faculties of reason and our faculties of observation do not cover all possible ways of reasoning or observing means that we require significant humility when faced with our own limitations. We must remember that statements of the form “It is inconceivable that X” is not, first and foremost, a statement about X, but a confession of the limitations of our own imagination.
Give the diversity of ways of knowing considered in this essay, and the rate at which humanity is even now just becoming aware of many of them, we would not rush to say that humanity, much less science, has a comprehensive grasp of all possible ways of knowing. Hume believed that he had all bases covered. We are less sure.
 Hume, David (1748, 2008). Enquiry Concerning Human Understanding. Oxford: Oxford University Press.
 As a general rule, the longer the piece of material is, the harder it is to keep the current flowing. The wider the piece of material is, the easier to keep the current flowing. The reasons for this are similar to why it is harder to pump water through a long thin pipe than through a short wide pipe.
If someone asks “How hard is it to pump water through this pipe?” You might reasonably ask, “How long is the pipe? And how wide?” Even to calculate the answer, you need to know observational data about the pipe in question.
 Even if I directly observe an ohmmeter read-out that says “2 milliohm”, the ohmmeter has encoded within it certain theoretical relationships between currents, voltages, and the display reading. Here is the technical account for anyone who cares. Otherwise, you can just go back to the main essay.
Resistance is a measure of how much effort it takes to keep an electrical current flowing through an object. If a current of 1 amp can be driven through an object by applying a potential difference of 1 volt across it, then it has a resistance of 1 ohm.
Resistance is measured using an ohmmeter. You connect the thing you want to measure to the terminals, and read off the resistance from the read-out scale. That said, there is a lot going on between the terminals and the read-out scale.
On the inside of the meter, the terminals are connected to a circuit which includes a voltage source, a resistance, and a coil of wire placed between the poles of a magnet. The coil has a pointer mounted on it, and is able to move against a spring. When you connect the terminals to the thing you want to measure, this creates a circuit through which the voltage source drives a current. The current in the coil of wire creates a magnetic field, which interacts with the permanent field of the magnets, and deflects the spring-mounted pointer. The scale is carefully marked to ensure that the pointer points to the value of whatever resistance is placed between the ohmmeter’s two terminals.
There is a theory about how potential differences, currents, and resistances are related to each other. There is a theory about how currents create magnetic fields, and how magnetic fields deflect spring-mounted pointers. When you ‘measure’ a resistance with an ohmmeter, what you observe is a pointer moving. You can only infer a resistance from this by invoking a chain of theoretical connections.
Of course, the way of measuring resistance described here is only one among many. But every single method requires a chain of theoretical connections between what you actually observe and any conclusion you draw about the the resistance of the object under consideration.
 Electrons moving through a material tend to bump into things. This means that you need to put effort in to keep the electrons moving, which means that the resistance is not zero. There are ways to stop electrons from bumping into things, usually requiring one or more of the following tricks: low temperatures, high pressures, and exotic materials. Theorists have no idea how you could get zero resistance at room temperature and pressure. After a lot of thinking, many people think it may be impossible.
 Indeed, the 1987 Nobel Prize in Physics was awarded to J. Georg Bednorz and K. Alexander Müller for their experimental work which demonstrated superconductivity under conditions that the existing theory said should not be possible. The exact mechanism for many superconducting materials is still a mystery.