Beyond Kant

This is a continuation of previous post on “The Knowledge of Childless Philosophers“. I would like to clarify some aspects of the theory of knowledge which have become muddled and confused because childless philosophers did not observe how children learn about the world, and acquire knowledge starting from scratch. If they had taken this as the basic model for how we acquire knowledge, they would have been able to avoid a huge number of mistakes. 

A realist methodology for science starts from the realization that scientific knowledge goes FAR BEYOND the realm of the observable. Electrons, Neutrons, Positrons have different kinds of charges, and act in different – incredible and amazing –  ways, but the link between them and observable phenomena is extremely weak and indirect. One of the readers’ comments on a previous post was: how can we learn about what we cannot observe? Paraphrase this to: how can we learn which of two slits a photon goes through, when we cannot see these events? This was exactly the question that Kant faced: How do we get knowledge which goes beyond what we can observe? It is obvious that rationalism will not provide an answer – if we start with self-evident axioms and use logic, we cannot ADD empirical information not already contained in the premises. It is also obvious that Hume’s empiricism will not work. Electrons are not there for us to see, and there is no amount of observations and experiments that we can show ordinary people which will allow them to deduce that electrons exists. Kant took ONE STEP beyond the realm of the observable to argue that our MIND supplies the (invisible structures) which organize the observations into a coherent and meaningful framework. This is what “transcendental” refers to – knowledge which transcends logic and observation. It is worth pausing to appreciate the value of Kant’s contribution. When Newton looked at the falling apple, HOW did he come to think of “gravity”? This knowledge is not there in the observation, because the same observation is routinely made by millions without thinking about gravity? In general, we can construct hypotheses about enormous spherical shells with stars pasted on them, rotating about the earth. In doing this, we are created models from our imagination, which do not correspond to observations, but serve as deeper explanations for what we observe. Kant realized that human knowledge of structures of reality was NOT based purely on logic and observations, as the rationalists and empiricist philosophers had argued. Of course, if he had observed children acquire knowledge, he would have learned this without so much difficulty. Also, seeing that children are experts at acquiring knowledge of the unobservables, he might have been inspired to write in simpler language.

My 3-year old daughter was noisily “helping” her aunt put her baby to sleep. To prevent her from waking up the baby, her aunt said “I think I hear your mother calling you”. My daughter raced out of the room to look for her mother. She found that her mother was engaged in conversation, and did not pay any attention to her coming into the room. She immediately deduced that she had been sent away from the baby and protested loudly that “My aunt sent me away from the baby” — looking through the appearances to arrive at the real cause why she had been told her mother was calling for her. Where did she acquire this knowledge, which was not part of what could be observed directly by her?

There are many examples of how children make inferences which are strongly in conflict with our imagined logics of scientific discovery. Children jump to generalizations from observing one fact, instead of patiently waiting to learn an entire collection and then deducing a law. In fact, as many psychologists have noted, children are born scientists. Three year olds learn difficult linguistic rules, and uncover hidden mechanisms at operation with ease. This is because, as Kant realized, we are born with mental structures which enable us to learn about the world we live in. There is strong evidence that emotions reflected by facial expressions are universal, so that knowledge of a range of human emotions is built into us. Babies differentiate between frowns and smiles and respond appropriately. Similarly, Noam Chomsky argued that the facility with which we learn languages shows that we are born with innate knowledge of grammatical structures. We are born with the capacity to imagine what the hidden structures of reality may be, and make good guesses about “Why” things happen.    

Where Kant went astray was in forbidding the cross-checking of these imagined mental structures with reality itself. Our mental models can be right – if they match reality – and wrong – if they don’t. Kant thought that it was impossible to check this – the true structure of reality could never be known because it was unobservable. While it is true that the mental structures we hypothesize to explain the observations can never be observed in external reality, this does not mean that there is no way to verify the existence of unobservable objects and effects. When we postulate the existence of electrons to explain some observable phenomena, we cannot go and look to see if there really are electrons. But we CAN use the hypothesis of existence of electrons to predict other phenomena that we would see if they existed. And when we see such phenomena, we get further confirmation of their existence. The empiricist REJECTION of this type of reasoning comes from the quest for certainty. Regardless of how many indirect tests we carry out, we cannot achieve the same level of certainty that we could from seeing and touching electors, and so if we confine our definition of knowledge to “Justified, True Belief”, we can never achieve knowledge about existence of electrons. If we forbid speculative talk (which was the intent of Hume and the empiricists) about objects which might exist, but about which we could never be certain, then effectively we bar talking about all subatomic particles and their properties. 

One of the key contributions of “Pragmatic” Philosophy is to give up on the quest for certainty, and settle for uncertainty in knowledge. Our entire life experience is based on navigating uncertainties, and making guesses about unobservables. Someone who sees a child growing and learning would never make the mistake of thinking that the child is acquiring mental structures of knowledge which do not correspond to structures of external reality. It would be crystal clear that the learning process involves use of hands, legs, and eyes, in addition to the brain. Furthermore, while it is clear that the child acquires new concepts – mental structures – it is also clear that increased success in navigating the world shows that these structures accurately reflect external reality.  

Instead of listening to childless philosophers, we should based our epistemology on our life-experiences in dealing with unobservables. Internal feelings in hearts of others about us are forever unobservable for us, yet the fabric of our social lives is woven from learning about and attempting to influence these feelings. Many posited unobservables have observable side-effects and implications. Many scientists were convinced of existence of atoms and molecules by Brownian motion, which could be observed, and could be explained by their presence. So to go beyond Kant, we must take our mental models seriously, as hypotheses about unobservable reality. These hypotheses often have observable implications. In fact, scientific experiments play a vital role in creating situations where the hypothesized objects and effects can be studied in isolation. Thus experiments play a crucial role in testing theories about unobservables. Such tests and experiments can never either decisively refute, or decisively  confirm, any hypothesis about unobservables. But as long as we learn to live with uncertainty, we can find near-refutations, or strong confirmations, of our scientific theories about unobservables. The Real Model – the correct unobservable true structures of external realities – is forever out of our reach, and known only to God. The progress of science, and our own life-experiences, show us that we can learn bits and pieces of it, and build on this learning, probing deeper and deeper into hidden layers of external reality.

Any study of scientific discovery will provide clear examples of how unobservable entities are discovered, and how we learn to manipulate them, even though we are unable to observe them. The discovery the certain liquids only combine in fixed proportions led to the hypothesis that these liquid were composed of molecules and that these molecules could only combine in certain fixed proportions. A diverse array experiments and phenomena could be explained by these same molecules with the same properties, giving weight to the hypothesis of their existence. An important point is that the molecular theory itself proved very fruitful as the basis of further, more complex and intricate hypotheses about the structure of matter. Those who were trained in the social product of scientific knowledge were able to make observations and experimentations of far greater sophistication than those who were not equipped with this knowledge.

Kant thought that our knowledge does transcend the bounds of observations and logic, but this transcendental knowledge is “ideal” – that is, it is projected by the mind onto reality, and it does not correspond to structures of external reality. We cannot know the true nature of unobservable external reality because we can never observe it. Advances in science substantially weakened the position taken by Kant. After Einstein’s discoveries of the counter-intuitive nature of curvature of time-space, it was not possible to maintain that time and space were projections of our mind onto external reality. Kant was also a prisoner of his time as major scientific discoveries of unobservable objects and effects had not yet taken place. It would be much more difficult to maintain that properties of atomic particles, electromagnetic phenomena, and astrophysics, are all projections of our minds, without correspondence to external reality. (For more detail, see Kant’s Blunder)

As opposed to Kant, Bhaskar Roy offers us “Transcendental Realism”. That is, we can have knowledge of objects and effects which we cannot observe with our sciences. To understand scientific progress, we must make a clear distinction between “Ontology” and “Epistemology” which is very muddled and confused in nominal (non-realist) philosophies of science. In fact, nominalist philosophies often commit the “epistemic fallacy” of supposing that if we cannot know something (an epistemological constraint), then it does not exist (an ontological conclusion). A weaker form of the epistemic fallacy, but with same consequences, is to say that if we cannot observe it, then it may or may not exist, but its existence does not matter for scientific theory. Bhaskar’s Critical Realist Philosophy of science is based on clearly distinguishing between ontology and epistemology.

Ontology refers to objects and effects in external reality. The existence of these objects and effects has nothing to do with whether or not we can learn about them. Bhaskar calls them the “Intransitive” portion of scientific knowledge.

Epistemology refers to HOW we can learn about objects and effects which are unobservable. This has a lot to do with our human capabilities – our hands and eyes and ears, and our abilities to manipulate our environment, conduct experiments, construct instruments, isolate causes, etc. This portion of our knowledge is social, and must be learnt and passed on – which is why Roy calls it the “Transitive” portion of scientific knowledge. The set of experiments, inferences, and reasoning, which leads us to know about the existence of electrons and their properties is very complex, and has been acquired over centuries. If this socialized knowledge was not transmitted to the next generation, electrons would continue to exist, but we would no longer know how we can learn about their existence. It is this knowledge which accumulates, though it is subject to errors and to major revisions from time to time, as Kuhn discovered. This is inherent in the nature of uncertainty associated with this knowledge.

In the next post we will consider how we can go beyond the observable data in econometric analysis to extract deeper information not present in the surface appearances, which are just the correlations.