What is the difference between knowledge and understanding?
The virtues of a new reliabilist approach...
The following is my entry into the Trinity College Cambridge Essay Competition.
The primary concern of this essay will be to explore the distinctions between knowledge and understanding, and to examine the role that an altered form of reliabilism plays in the process of obtaining either state. In essence, this essay will argue that understanding is related to–yet also subtly different from–the reliabilist method of attaining knowledge. I argue that scrutinising traditional reliabilism thus clarifies the knowledge-understanding distinction. However, I first offer an analysis of a key distinction often omitted, that of materialism versus normativity. I argue that it is in between this conceptual space that we can most clearly comprehend the differences between knowledge and understanding.
Materialism & Normativity
A materialist view of knowledge can be demonstrated with reference to the following proposition:
James knows the way to Tipperary.
Whether James can reliably guide us or not, whether he has acquired this trait by luck or not, or whether it is an ability, acquaintance, or propositional form of knowledge, is all to propose that James' brain is in a certain state when we say that he knows the way to Tipperary (Chomsky, 1992). This is the materialist use of the verb to know.
The normative use can be illustrated as follows:
Tim: I know the Earth is flat.
Prof Dawkins: There's no way you can possibly know that.
In this passage, Prof Dawkins is illustrating that a brain state is not a sufficient condition for knowledge as common parlance defines it. The main reason traditionally given for why Tim cannot know the Earth is flat is because it makes a mockery of language to say that we are able to possess knowledge of untruths, a view dating back to Plato (McDowell, 1987). The normative view(s) of knowledge thus impose external sets of criteria, like truth (but also, most famously, justification and belief). This Socratic pursuit of necessary and sufficient conditions comes under scrutiny in Gettier cases, perplexing edge cases involving luck and the accidental acquisition of correct answers.
Whilst these cases are problems for the normativist view of knowledge, the materialist view has its own problem. Namely, neuroscience remains, as of yet, indifferent to any external conceptual imposition lacking scientific validity. This is to say, no matter how good brain scanning technology becomes, it will never be sufficient to simply examine a brain scan (a descriptive account of what is) in order to answer a question about the possession of a socially constructed definition of knowledge. (By socially constructed, I am referring to the criteria found in something like the JTB definition).
Likewise, the concept of understanding has both materialist and normative aspects. There are, however, some important differences. At the level of the brain, comprehension is often simply an extension of knowledge. For example, John may know that a certain phrase is not grammatically correct because he has learned the phrase through association. John might be able to recognise the incorrect use without failure, but if he does not know the rules of grammar he cannot be said to understand why the phrase “we done it” is incorrect. To achieve understanding, one must acquire the extra knowledge required for grammatical comprehension.
However, note a subtle distinction: Perhaps John might be more effective at spotting incorrect grammar—despite his lack of understanding—than an individual who understands the rules, but suffers from some type of nervous disposition that affects his pattern recognition. We’ll call this odd individual Fred. When we ask Fred to explain the rules of grammar, he does so perfectly. But under test conditions, John beats Fred every time. Is John’s result luck? If John is not guessing, but relying on some type of acquired instinct after years of familiarising himself with the appropriate use of “done” versus “did”, then this is not luck in any normal sense–there is no arbitrary, random element. Hence, understanding cannot solely be about efficacy. Understanding is the ability to know why something is the way it is. Here, however, is where normativism must rear its head.
The phrase why something is the way it is requires normative input. To see why, take the example of understanding E = mc². The best physicist will, in his answer, regress all the way to the origins of the universe in explaining Einstein’s equation. If his hippocampus and prefrontal cortex are in a certain state (which is to say he is not just reading a textbook aloud, but has encoded long-term memories), it is safe to say the physicist understands why E = mc².
But such a regression (to the universe’s origins) can be made with any question. Yet society clearly sets much lower thresholds. Nobody thinks that an art historian does not understand Vermeer because she does not understand the electromagnetic system. Again we are left in the position of saying that, by itself, neuroscience will never answer the question of whether understanding has been achieved. Reference to social norms is unavoidable.
In sum, fact and fiction can be indistinguishable at the level of the brain, as can understanding and the lack of it (depending on the normative threshold). If we wish to maintain that it is never possible for Tim to know the Earth is flat, even though he could have a material (brain state) understanding of this fiction after binge-watching YouTube conspiracy videos, a compromise must be sought between normativity and materialism. I argue that this task requires the introduction of a middleman, delivered by what I have called critical reliabilism.
Critical Reliabilism
Standard reliabilism relies on an implicit truth-ratio to conclude whether or not knowledge has been attained. The truth-ratio is the percentage indicating how frequently a reliable method brings forth a true result and, thus, knowledge. Yet reliabilism ostensibly makes no effort to incorporate the role of understanding (Goldman, A. and Beddor, B. 2016). Despite the many refinements to the reliabilist theory that have been developed, the definition of what constitutes a justifiable reliable process is somewhat arbitrary, so that it becomes impossible to objectively determine whether a knowledge-seeking method that possesses a truth-ratio of 0.7 (or some other arbitrary percentage between 0.51 and 1) may or may not be considered a sufficiently reliable process. If we wish to understand knowledge and understanding from a pragmatic perspective, we need critical reliabilism.
C-reliabilism holds that one requires a critical cognitive capacity. Some may argue that this cognitive capacity is not needed because the presence of a sufficiently reliable truth-ratio is indicative of a reasoning process. Let us return to the example of grammar to show why this supposition is misguided. According to a reliabilist definition, a person who has learned grammatical distinctions through exposure is capable of learning that ‘we done it’ is an example of poor grammar, because a sufficient truth-ratio has been achieved. The c-reliabilist doubts this by incorporating understanding. To understand requires possession of a reason as to why another belief about a topic cannot be true, or why it must be unlikely. Once one has gained experience that can support a reason to suppose why one phrase is correct, and why another belief on the same topic must/is likely to be false, then one can be said to possess understanding according to c-reliabilism.
For example, if the person who learnt the above grammatical distinction were a c-reliabilist, they would claim that they have not understood the distinction, because they cannot explain what would be wrong about the phrase ‘we done it’. The c-reliabilist definition satisfies both the materialist requirements of understanding and the normative requirements. It manages to satisfy the materialist accounts by associating understanding with the possession of a brain state that is distinct from the brain state of knowing. That is because someone such as the conspiracy theorist mentioned previously cannot possess the brain state of rebutting a counterfactual (the counterfactual: the earth is flat) whilst still believing the counterfactual.
C-reliabilism manages to satisfy the normative accounts by explaining that understanding does have a distinct definition from knowledge, for we can still use traditional reliabilist methods to satisfy our everyday normative use of knowledge. Harkening back to the case of the art historian, we can say that the art historian understands art history on both a normative and material level, for she can explain why a sculpture from Imperial Rome is distinct from that of Republican Rome, by explaining how the latter could not be of the same era as the former, and by giving her explanation she experiences a brain state which could not have been experienced without an adequate (from a critical reliabilist view) reason for the distinction between sculpting styles. But how much detail must our historian include in her counterfactual to warrant the label of understanding?
I previously noted that the boundary conditions on the counterfactual inevitably derive from social norms, so it is worth clarifying what they should be. In brief, we might say that as long as our art historian offers a set of conditions for the falsification of her knowledge, this would indicate some level of understanding for the critical reliabilist. However, to achieve full understanding, the art historian needs a materialist level of understanding in addition to the normative. This involves possessing a reason to justify why a counterfactual is wrong, which, if it were a faulty reason, would lead to the destruction of some of our most basic assumptions about reality.
We may choose to subject the distinction between Imperial and Republican era Roman statues to this point by stating that we only have a sufficient justification that the two statues are from distinct eras if some property of the statues, such as their atomic structure, is distinct. To deny this fact would require a denial of some basic assumption about the nature of the universe, namely the identity of indiscernibles (Forrest 2016). If the two statues have a differing atomic structure, this indicates that they are from different ages and cannot be the same sculpture. Therefore, to deny that they are different would mean either to deny the identity of indiscernibles or to deny the means by which we have determined they have a different structure. To deny the latter would imply a denial of the former, as we have formed our empirical methods on a priori deductions.
This brings us neatly to the normativity of knowledge and understanding. It should be clear that c-reliablism accords with the view that understanding and knowledge are differentiated by what one can do with each (De Regt 2004, 2017; Wilkenfeld 2013; Elgin 1996). In this view, one who is unable to use their knowledge to learn anything new would be considered lacking in knowledge, insofar as such an individual may not be capable of making any deductions from their ‘knowledge’ of actions. In fact, we can doubt whether such an agent possesses knowledge of a topic in the first place, as we don't know whether they have any means of distinguishing concepts from each other, a capacity which all epistemologists agree is required for knowledge. Insofar as knowledge is already considered valuable, instances of understanding are associated with the capacity to achieve greater knowledge, so understanding can be said to be more valuable than knowledge. (Kvanvig 2003).
Conclusion
In sum, the cash value of critical reliabilism is more than simply an insight into the knowledge-understanding distinction. In clarifying this distinction, critical reliabilism offers an answer to a deeper question: what it means to be truly educated. That it must mean more than knowledge acquisition should be obvious. I might be the best at identifying Roman marbles or incorrect grammar, but if I lack (a) the deeper knowledge of the system that generates these outputs and (b) the metacognitive counterfactual, I am without something of immense utility: understanding. Thus, following on from Kvanvig’s view that understanding is more valuable than knowledge, we are now well positioned to add an additional reason why it is so. Understanding is the load-bearing unit of knowledge. Without it, society stagnates.
Bibliography:
Chomsky, N., 1992. Professor Noam Chomsky in conversation with University of Oviedo students, 1992. [online] Available at: (Timestamp: 1:06:30) [Accessed 31 May 2022].
de Regt, Henk W., 2004, “Discussion Note: Making Sense of Understanding”, Philosophy of Science, 71(1): 98–109. doi:10.1086/381415
de Regt, Henk W. and Dennis Dieks, 2005, “A Contextual Approach to Scientific Understanding”, Synthese 144(1): 137–170. doi: 10.1007/s11229-005-5000-4
de Regt, Henk W., Sabine Leonelli, and Kai Eiger (eds.), 2009, Scientific Understanding: Philosophical Perspectives, Pittsburgh, PA: University of Pittsburgh Press.
Elgin, Catherine Z., 1996, Considered Judgment, Princeton, NJ: Princeton University Press.
Forrest, P. (2016). The Identity of Indiscernibles. Winter 2016 ed. [online] Stanford Encyclopedia of Philosophy available at: https://plato.stanford.edu/entries/identity-indiscernible/
Goldman, A. and Beddor, B. (2016). Reliabilist Epistemology. Winter 2016 ed. [online] Stanford Encyclopedia of Philosophy. Available at: https://plato.stanford.edu/entries/reliabilism/.
Kvanvig, Jonathan L., 2003, The Value of Knowledge and the Pursuit of Understanding, New York: Cambridge University Press. doi:10.1017/CBO9780511498909
Plato and Mcdowell, J. (1987). Theaetetus. Oxford Clarendon.
Wilkenfeld, Daniel A., 2013, “Understanding as Representation Manipulability”, Synthese, 190(6): 997–1016. doi:10.1007/s11229-011-0055-x