PHI 162: “The Problem of Scientific Knowledge through Demonstration” (according to Aristotle)

Jolene Patricia Brown

Dr. M.V. Wedin

PHI 162: Aristotle

27 April 2010

The Problem of Scientific Knowledge through Demonstration

Aristotle claims that scientific knowledge can only arise in the situation that one can demonstrate what they know by presenting the knowledge in a coherent form with specific rules. In this way, his definition of scientific knowledge can be compared to a syllogism: it takes two true premises, argued together, to be able to claim new true information in the conclusion of the argument in the form of the syllogism. Like a syllogism, scientific knowledge consists of true claims that one must understand, and have seen demonstrated, taken together in an argument to be a conclusion that consists of scientific knowledge. He writes:

…it is necessary for demonstrative understanding in particular to depend on things which are true and primitive and immediate and more familiar than and prior to and explanatory of the conclusion… (71b.20)

In the above quote he states four dependencies that must be necessary for understanding of scientific knowledge to occur: a premise must be true, primitive and immediate, more familiar and prior and explanatory. These dependences are important because without them a conclusion would suffer from inconsistency, false claims, and a lack of demonstration that would be necessary for knowledge to be obtained.

Each of the four dependencies represents some form of demonstration in and of itself. The truth dependency, the first one on his list, is that which claims that the premises must be true in order for the conclusion to be true: truth in the premises demonstrates the relationship of truth in the premises to the truth contained in the conclusion, rendering the information in the conclusions as true. The second dependency, primitive and intermediate, refers to the relation of the premises to the conclusion for the information in the premises must precede the information in the conclusion or otherwise entail the conclusion. It would not be possible for a premise that that contains information that is, for instance, chronologically later than the information contained in the conclusion. This dependency demonstrates the relationship of the information in the premises to the information contained in the conclusion and the importance of entailment and logical following as going from premise to conclusion in the course of the argument. The third dependency is more familiar and prior which refers to the relationship between the information and the individual who knows it. One must be familiar with the information contained in the premises in order to use it to make a claim about that information in a new way, such as in a conclusion. “Being prior and more familiar” is that which is “nearer to perception”—something that can been demonstrated to the individual in such a way as to produce knowledge, which can in turn be used as premises to form even more knowledge in the conclusion (72a.1-5). The last dependency is that the premises must be explanatory, meaning that the information contained in the premises must explain the information contained within the conclusion, thereby demonstrating the connection between the relevance of the information in the argument.

The problem with Aristotle’s claim of demonstration as a means of obtaining scientific knowledge is that there is a regress of circularity that occurs in the rigorousness of providing demonstration for all knowledge beginning with the premises that precede the conclusion and extending back to the “first premises.” If Aristotle is correct, there will be no “first claims” that would be found underlying all other claims leading to a conclusion: if there were it would create a problem because either those first claims would not meet the criteria of the dependencies, or one would have to continue their demonstration in a circular argument that would quickly become non-demonstrable. Aristotle makes an argument against the circularity of knowledge by stating that “if demonstration must depend on what is prior and more familiar…it is impossible for the same things at the same time to be prior and posterior to the same things” (72b.25-30). We must remember that being “prior and more familiar” demonstrates the relationship of the information to the individual, and in this way Aristotle is correct: for an argument to be circular, then the individual would both have to know and not know the premises simultaneously in order for the argument to complete itself in a reciprocal fashion. Since the ability to simultaneously know and not know is impossible, the idea that all arguments are circular is not an option.

Instead of pressing for more justification of demonstration as the only means of scientific knowledge, Aristotle then explores the idea that the immediate, or that the dependency between the premises and the conclusion, is not necessarily demonstrable and that instead knowledge at a certain point can be obtained through familiarity by definition (72b.20-25). Aristotle earlier defines “definition” as a “posit…but not a supposition” (72a.20), meaning that unlike a supposition that “assumes either of the parts of a contradiction”—like an assumption—a definition gives information that one must assume but that does not nullify either part of a contradiction. Instead, a definition outlines the meaning of a premise in such a way that it maintains truth, and allows the formation of new information in the conclusion thereby creating new premises for argument: the first “first claims” needed to source scientific knowledge. In claiming that it is possible for some premises, namely those that are immediate, to originate from definitions, Aristotle almost convinces the reader of the definition as a solution to the problem of infinite regress, but even that falls short. The problem with the idea of definition is thus stated:

Since one should both be convinced of and know the object by having a deduction of the sort we call a demonstration, and since this is the case when these things on which the deduction depends are the case, it is necessary not only to be already aware of the primitives…but actually be better aware of them. (72a.25)

By this explanation it is not sufficient that a definition might allow a premise to exist as a primitive, since it is necessary for an individual to be already aware of the premises, but “better aware” of them than the conclusion itself. Since it would be hard to take the definition of a very abstract idea and construct an argument such that the conclusion would not only include information from both definitions, but also create a new idea that is concrete seems impossible. A definition would need to be so well-known by the individual that it would become another premise that would need another demonstration in order to prove all the points of the definition in such a way as to create a conclusion, leading back to the original circular pattern of demonstration needed for scientific knowledge.

Unhappy with the options he presents about the source of knowledge Aristotle instead rejects all accounts, and claims there must be another way for one to achieve knowledge. His options as they stand are to concede to circular reasoning, resort skepticism, or have an infinite explanation that no one would ever be able to conceive. None of these options present a case for scientific knowledge as he would like to accept—in demonstration—so instead of claiming one over the others, he rejects all of them as false, and by doing so implies the option of something unknown even to him: “…hence, since there are few such things in demonstrations, it is evident that it is both empty and impossible to say that demonstration is reciprocal and that because of this there can be demonstration of everything” (73a.15-20). By rejecting all those things he denies are the sources of truth he does not explicitly state what it is that is the source of truth, and instead makes an oblique claim that there is something else that is unknown at this point that is the source. Scientific knowledge, therefore, can be had, but we reach a point in demonstrating the truth of it that knowledge itself remains source less.

PHI 102: "Indexical Contextualism as a Solution to Dretske's 'Zoo Problem'"

Indexical Contextualism as a Solution to Dretske’s “Zoo Problem”

Epistemological contexualists argue that the variation of standards in knowledge is what guides an attributor to assign to a subject knowledge or ignorance. One example of this kind of variation in standards is Dretske’s “zoo case,” or an alternative case presented in class by Dr. Mattey of the people in different situations waiting in the airport for a flight. The similarity between both of these cases is that a subject is attributed knowledge in one context, and ignorance in another, with no changes in the situation of the subject, but changes in the context of the attributor. This unusual occurrence creates a lack of argument closure that undermines the original assessment of the attributor leading to a reassessment on the part of the attributor. This lack of closure lends itself to a solution found within indexical contextualism in that it is the context of attributor that confirms the knowledge or ignorance of the subject.

An example of this problem can be formed through a modification of a scenario given by Dr. G.J. Mattey early in the fall 2008 quarter in his epistemology course (PHI 102). In one discussion the question was raised as to whether or not one knows if an old car is a genuine antique, or if it is a car built from a kit to look like an antique, and, he argued, if one has not the training to identify the true antique cars from the new kit cars, then one may not say they “know.” To retool this scenario for the purpose of contextualism, suppose there is a museum of antiquated cars, and subject S is looking at one of the cars on display, and he reads a sign that says the car was built in the year 1945. An attributor A is looking on when S tells A that S knows that the car was built in 1945. S acquired this information, he tells A, by reading the sign on the wall next to the car stating that the car was built in 1945. A has evidence that S knows that p such that the sign was supplied by a reputable source (a museum), and S believes that the car was built during 1945 from some basic understanding of what cars built in that era look like. This casual scenario, context 1 at time 1 (C1 at T1), allows A to draw with some confidence the conclusion that S does, in fact, know that p (the car was built in 1945). This argument can be written:

A attributes knowledge to S at T1 in context C1, given the casual nature of this scenario.

To change the scenario such that the attributor A has a more pressing situation to consider may change the attribution for knowledge to ignorance on the part of subject S. Perhaps a camera crew walks over to A after S claims knowledge and tells him that if S is correct in his knowledge that the car was built in 1945, then A will win a million dollar prize. S is not told about this camera crew, and has no idea there is a contest going on in the museum. In this way, S’s context has not changed: he still believes that the car was built in 1945, and has based this belief on the sign posted in the museum. A’s context has changed with this information and A might then be even more willing to attribute knowledge to S in this scenario given his enthusiasm for winning a substantial prize. A is then told by the camera crew that several of the cars in the museum have been replaced with forgeries of antique cars built from kits, nearly indistinguishable from the antique cars they represent and none of the forgeries were built in 1945. A’s context has changed yet again and A must then decide if he is going to attribute knowledge or ignorance to subject S, and based on his correct choice will be awarded the prize money. A, in this less casual context, may be less likely to attribute knowledge to S simply based on the information that A has that some of the cars in the museum he now knows to be forgeries. Thus, we have the following propositions:

1. A at C1 believes that S knows at C1 (at T1) that the car in question was built in 1945.
2. A at C2 believes that S does not know in C2 that the car in question was not built in a year other than 1945.
3. A at C2 believes that S does not know in C2 that the car in question was built in the year 1945.

Given this change in circumstances, or context, it would seem that A has now the potential to change his original attribution of knowledge to ignorance, and would be justified in doing so, with the intention of winning the prize. Proposition one was correct in the context in which A attributed knowledge to S because he did not have evidence otherwise to attribute ignorance instead. In light of new evidence, A was then justified to attribute ignorance instead of knowledge to S, if he doubts S’s ability to distinguish antique cars from kit cars. This change in context leading to a new attribution is what causes problems in epistemology, which could be solved through indexical contextualism.

The indexical contextualist would argue that the word ‘knows’ is indexical, that is, the word ‘knows’ in one context (C1) is being used in a different way than in another context (C2). This seems to be a good solution to this problem of contextual difference that we see in this problem. If one views the word ‘knows’ as indexical then it makes sense that the word will mean one thing in one context and another thing in another context. As Steven Luper, from the Stanford Encyclopedia of Philosophy, writes:

According to contexualists, whether it is correct for a judge to attribute knowledge to someone depends on that judge’s context, and the standards of knowledge differ from context to context…What passes for knowledge in ordinary contexts does not qualify for knowledge in contexts where heightened criteria apply (Luper).

It is A’s change in context, in the form of being monetarily inclined to make the correct decision in attributing knowledge or ignorance to subject S, that creates a problem with closure in the argument of whether or not S can be attributed knowledge. Contextualism can salvage closure by allowing the use of the word knows to change from context to context. Returning to our earlier problem, A was not wrong in C1, when he did not have anything at stake by attributing knowledge to S, despite his later evidence stating that some of the cars in the museum were forgeries. Since A did not have that evidence at the time he made the utterance, then at the time he made the utterance his attribution of knowledge was correct.

One solution that seems to work well in cases such as these is Stewart Cohen’s Indexical Contextualism. Cohen’s attempt to solve this problem holds that closure should be kept intact. To quote Mattey’s lecture notes: “Indexicality is certainly one way to generate sentences which express propositions which vary with the context of assertions,” and Mattey goes on to say “one way of expressing the indexical view is to think of ‘knows’ as an elliptical verb that contains a hidden parameter” (2008). For example Cohen argues that the word knows works like other indexical words in the language, for example the word “I.” When the sentence “I am hungry” is uttered by a person, the utterance at the moment it is uttered is true for the person making the utterance (as long as the person is making the utterance in sound mind, and not because the person is suffering from a mental condition or some other mental disturbance). This flexibility allows for the argument to draw closure from the propositions within the argument being evaluated at the same context, and drawing conclusions of knowledge from within those contexts. Mattey discusses how Cohen’s indexical contextualism attempts to salvage closure within his Philosophy 102 web notes:

Cohen holds that closure can be saved by allowing that the sense of ‘knows’ changes from context to context. Within the same context, closure holds, but when the context on the premise and the conclusion operate with a different sense of ‘know’ due to shifting contexts of attribution, there is no failure of closure (2008).

It is important to note the consistency being attained by looking at the antecedents and consequents of an argument in light of the same context, as pointed out in the above quotation. Take, for example, our case earlier, of the attributor and the subject in the car museum. It would be unlikely that any reasonable person would agree that A was incorrect at the time T1 in attributing knowledge to subject S. The evidence that was available to both the subject and to the attributor when he made the attribution of knowledge seemed, at that time and in that context, sufficient enough to warrant the attribution that S had knowledge of the car being built in 1945. It was only after the introduction of new relevant evidence that A’s knowledge attribution is then doubted, and A becomes unlikely to attribute knowledge to S, and even then would only be doing so under the context of a higher-stake situation in his favor (namely, monetary gain).

Based on this and other examples there does not seem to be a “fixed” standard of knowledge. Instead, knowledge is something that is determined by context, and we can accomplish this while still keeping closure intact through Cohen’s explanation of the indexical way that the word ‘knows’ works. Without a fixed standard of knowledge there is a flexibility in the truth and falsehood of sentences, as they are uttered by the subject and the attributor, and the context in which the sentence is uttered.

Works Cited

Luper, Steven, "The Epistemic Closure Principle", The Stanford Encyclopedia of Philosophy (Winter 2008 Edition), Edward N. Zalta (ed.), forthcoming URL = .

Mattey, G.J., “Contemporary Epistemology III: Language.” Version 1.1, November 9, 2008.