Francis Remedios
Institute of Philosophy
University of
Leuven
Kardinaal Mercierplein 2
Leuven, B-3000
Belgium
e-mail: francisr@shaw.ca
published in Philosophical Inquiry, Vol. XII, Winter-Spring, 1990 No. 1-2, pp. 44-58
Introduction
The choice of a theory of
justification is central to epistemology with the main contenders being
coherentism and foundationalism. To the coherentist the problem for the
foundationalist is: what is the source or rationale of the non- inferential
warrant which is supposed to attach to a basic proposition? In other words, are
there epistemic principles (principles which state that all beliefs with a
certain feature are or are not justified) which either justify basic
propositions or guarantee their truth? That is, is there a stronger connection
between justification and truth other than the trivial connection that a person
is justified in believing P only if the person is justified in believing that P
is true? The charge can be stated this way:
1. For any proposition to be justified, there must be principles which govern justification of that proposition. 2. But those principles also have to be justified. 3. Foundationalist epistemic principles arc not justified because they are not internally connected to truth via a mental state.4. Hence, they are just 'if, then' principles in which no
mental is involved.
The
coherentist claims that epistemic principles can be justified within a
coherentist system of beliefs because warrant for propositions can be derived
from other propositions. In other words, epistemic principles are justified by
the overall system of beliefs; that is, the whole system justifies the
principle.
According to the coherentist one must know epistemic principles
are true if knowledge or justified belief is governed according to them. L.
Bonjour, a coherentist, argues that no beliefs are justified independently of
the relations they bear to other beliefs, so that no beliefs are
'foundationally' justified:
l. For any belief B, if B is justified for S, then S must have a reason to think B likely to be true. 2. If S has a reason to think that B Is likely to be true then this reason can only consist in the justified beliefs that (a) B has some property G, and (b) beliefs having G are likely to be true. 3. But if the reason for B's being justified for S consists in the justified beliefs (a) & (b), then B is not foundationally justified. 4. Therefore, no belief is foundationally justified.
It is important to note that Bonjour's argument does not simply involve the claim that if some belief is justified, then some other beliefs are justified. Rather, his view is that a belief B is justified for S in virtue of the justified beliefs that B has property G and that beliefs having G are likely to be true. Bonjour is led to this view in answer to the question, "How can a person be justified in accepting a contingent belief, if he does not believe a fortiori, does not know anything about it which makes it all likely to be true?"(1) In other words. a coherentist requires that he knows the property that make most beliefs true at the outset of the process of justification. This is a version of the doxastic ascent argument,(2) for Bonjour is asking the meta- question: "What warrant does a belief require to be immediately justified if one does not know the principle to be true?" This last question leads to the further question: "How does one know the belief to be true?" The question the coherentist asks is: "What is the foundationalist's criterion which distinguishes true basic beliefs from false basic beliefs if one can be justified without knowing the principles of how one is justified?"
In this paper I shall give the foundationalist's response, who claims that the problem of the criterion is not an insurmountable issue for him. From our basic knowledge of many things we can establish criteria, and notwithstanding, Bonjour we do not need a criterion for basic propositions from which epistemic principles can be established. The coherentist is begging the question by demanding that fixing a criterion be primary. Initially, I shall provide a brief history and a schemata for foundationalism. After that Firth's account of the epistemic priority of basic propositions will be examined. Then I will explicate Sosa's discussion of the common structure of foundationalism and coherentism and of the argument that coherentism cannot maintain coherence as a non-epistemic property without going to a regress. I shall argue that Van Cleve's account of how epistemic principles are justified without circularity is correct and that it meets Lehrer's demand that a theory of justification should be explanatory. Finally, I shall discuss the immediate justification of epistemic principles in terms of the Chisholm/Sellars debate on a priori knowledge and mediate justification of epistemic principles with respect to a conceptual framework.
1.
A Brief History of Foundationalism
The history of
foundationalism can be traced back to Aristotle's remarks on scientific
knowledge in the Posterior Analytics, Bk. l, chap. 2 & 3, Aristotle argues that because we certainly do have demonstrative knowledge which is the
product of inference from other known premises, we must also have
non-demonstrative knowledge which is not the product of inference from other
known premises.(3) Aristotle reasons that this has to be so because, if one is
to know the conclusion of a demonstration, one must know first the premises of
the demonstration, and if the premises of any demonstration are knowable only if
always demonstrable by appeal to other known propositions as evidence, then
knowing the conclusion of any demonstration would require passing through an
infinite series of demonstrable premises. Hence, since we cannot pass through an
infinite series and because we do have demonstrative knowledge, there must be
some premises which are not demonstrate by appeal to other known premises.
Demonstrative (non-basic) knowledge must ultimately rest on basic foundation
which are known but not inferred from other premises, basic knowledge may be
called foundational because it provides the evidential base without which there
could be no non-basic 'knowledge' and from which we generate the known evidence
justified for ill demonstrative knowledge. Aristotle's point is that basic
knowledge does not require the satisfaction of any evidence condition or
criterion, non-basic knowledge does require the satisfaction of some evidence
condition. It was Descartes, however, who anticipated foundationalism in its
modern form, by arguing that knowledge required acceptance of indubitable claims
and offering a description of the content of these claim:. On the other hand
empiricists thought that self-evident position were empirical propositions.(4)
The goal of foundationalism is to produce an account of the structure of
empirical knowledge in general and the motivation is to end the regress of
justification. A number of metaphors have been used to explain foundationalism.
We are asked to picture human knowledge as an edifice, wall, or inverted
pyramid, the foundation of which supports all the floors above, but they do not
support the foundation.
The structure of an individual's beliefs at a given time provides a natural starting point for foundationalism. This structure can be viewed as a system composed of individual's belief set at the time, and various evidential relations that obtain between members of the set. The belief set of any subject S at any time can be defined as the set of propositions that S believes or accepts as true at t. Foundationalists, of course, are primarily interested in two sorts of propositions in this set, those which arc candidates for the foundational and the candidates for the non- foundational. The former are called basic propositions and the latter are called non-basic propositions. A definition of a basic proposition is: a proposition p is basic for a subject S at time t if, and only if, (i) S does not require any propositions Q as evidence for p (ii) p is a member of S's belief set at t, (iii) p provides evidence for some other proposition q which is also a member of S's belief set at t.
Contemporary foundationalists do not hold that no account can be given of why basic propositions are justified. Rather, they repudiate a specific kind of justification condition. Basic propositions, they say, must satisfy some justification condition but the satisfaction is not to be sought by the way of derivation or support by other propositions. So, they urge, the basic proposition satisfy a justification condition in virtue of being immediately justified. Various analyses, for example, contend that self-evident propositions or self-warranting propositions satisfy the justification condition.(5) The constant theme is that basic propositions end the infinite regress of justification, for they are non-verified and non-supported, they are the foundation upon which the person's belief structure stands.
II.
Epistemic Priority
An example of why basic propositions
in the foundation of knowledge be elucidated by Chisholm's discussion of
Socratic questions. For Chisholm, Socratic questions are not "How do you known?"
questions or "why do you believe?" questions, rather they are questions like
"What is my justification for thinking that I know this?" They need not be asked
in a sceptical way, but in a way that looks for an analysis of what is already
accepted or recognised. A sample answer to a Socratic question looks like this:
What justifies me in counting it as evident that he has that disorder is the fact that it isevident that he has those symptoms and if it is evident that
he has those symptoms it is evident that he has that disorder.(6)
For Chisholm, there is a proper stopping place for Socratic questions. viz. basic propositions or what he calls self-presenting states of affairs. For example, "I seem to have a headache", if the state of affairs of seeming to have a headache occurs; then it is necessarily evident to me. Chisholm's version of basic propositions as the proper stopping place for Socratic questions is characteristic of how most versions foundationalism view the place of basic propositions.
I will look at two versions of foundationalism that of Firth and that of Lewis. Firth notes that some statements have some degree of warrant which is independent of (and in this sense prior to) the warrant (if any) that they derive from their coherence with other statements".(7) There are many versions of this thesis one might hold. C.I. Lewis, for instance, holds that sub a non-inferentially warranted statement is one. which characterizes the content of my present sense experience. Lewis holds that such statements are certain and that no other statement at the time is more warranted for that person birth accepts the view that the only inferentially warranted statements are about one's own experience, however, he suggests that the thesis of epistemic priority does not commit one to the view that these statements are certain. There are weaker degrees of warrant which one might attribute to these statements.
Firth(8) has three variants of self-warrant: the first maintains that the justification of a basic proposition may he increased when it receives support from other propositions, e.g. by cohering with them, and decreased when it fails to receive support, but in no cue is its warrant ever to the point where a person would be more justified in disbelieving the proposition than believing it. Second, it might be assumed that a basic proposition has some initial justification which is not be feasible through failure to receive support from other propositions nevertheless, the degree of initial justification is not great enough to insure that a person who accepts the proposition is always more justified in believing than in disbelieving it. Thirdly, one could adopt the second view with the exception that even the initial justification of basic propositions is held to be defeasible. While each of these alternatives provides a plausible account of non-inferential warrant the first account is the most plausible account of non- inferential warrant of the three because it claims that the basic propositions are basic but justification for the proposition is increased by coherence with other statements. Notice that this is different from the coherence theory of justification which holds that there are no basic statements. The second and the third positions are too weak since the foundationalist demands that basic propositions do not require any justification other than immediate justification. On the second and third views the degree of initial justification is not great enough for immediate justification in the intended sense.
IlI.
The Structure of Foundationalism
So far I have remained
neutral with respect to foundationalism which can be characterised either
formally as a structural notion of substantively as non-epistemic notion with,
respect to a person's belief set. Now I shall explicate "formal
foundationalism". Formal foundationalism can be held to be the study of
normative or evaluative principles of any sort, for example, those of ethics.
Formal foundationalism in ethics tries to fix the goodness of events or states, or the rightness of actions, recursively. A simple utilitarian theory says this:
l. Every event of someone undergoing pleasure is
good.2. Every event that causes a good event is good? and
3. Every event that is good is so in virtue of (1) or (2).
Analogously, formal foundationalism in epistemology would say that:
l. Every belief with a certain non-epistemic property
F is justified,2. If a belief bears relation R to a set of justified
beliefs, then it is itself justified, and3. Every belief that is justified is so in virtue of (1)
or (2).(9)
Formal foundationalism in epistemology holds that the notion of epistemic justification or warrant is recursively definable in terms of a non-epistemic basis and non-epistemic generator; that is, there is a special set which are the axiomatic starter set of truths, and a grounding process for validating truth claims in terms of others. The main argument for formal foundationalism is that the evaluative features arc generally supervenient or consequential generally, and hence also the feature of epistemic justification.(10) For example, an apple may be a good apple in virtue of having certain non-evaluative properties: in virtue, let us say of being sweet, juicy, large, etc. If so, then its evaluative property of being a good apple "supervenes" upon its complex of non-evaluative properties that includes being sweet, juicy, large, etc. Hence, any apple which is sweet, juicy, and large can be called a good apple. In epistemology a belief is justified in virtue of supervenience of justification over non- epistemic properties and any belief which has such and such non- epistemic properties is justified. According to the doctrine of supervenience, evaluative and normative properties always supervene on non-evaluative non-normative properties.
Contrary to popular opinion the coherence theory in epistemology is also a kind of "formal foundationalism" as Sosa correctly claims; that is, coherentism is structurally the same as foundationalism in that it also tries to provide principles that specify the conditions within which beliefs are justified. Thus a coherentist chooses coherence within a set of beliefs of a certain sort as a basis and deduction as the generator. For example, a coherentist may had that a belief is justified if and only if either it coheres within a comprehensive and diverse set of beliefs held by the subject or it is deduced by the subject from a set of such beliefs. In fact, the coherentist usually has an all- encompassing basis that absorbs all generators, but this is quite compatible with formal foundationalism. Coherentism is a limiting ease of formal foundationalism because coherence (explanatory relations, relations of probability or logic) is the basis which encompasses deduction. Coherentism is opposed not to "formal foundationalism, but to the view I have specified as epistemological substantive foundationalism; the conflict between them is over what basis property to choose in the basis clause of the recursive definition of justification. Foundationalists choose some form of basic propositions whether it be from a privileged access or a causal process account which holds to a nomological connection between the knower and truth, and coherentists choose coherence. What Sosa correctly points out is that the coherentist and the substantive foundationalist share a primary goal, the development of a formal foundationalist theory of the highest grade for they both want a simple theory that explains precisely how epistemic justification supervenes, in general. on the non- epistemic But this should not blur us to the critical differences between them.(11)
Sosa's(12) argues that the problem with coherentism is that if coherence is valid, then, supervenience over the non-epistemic must be given up by the coherentist. If the coherence theory is to accept the principle of supervenience of justification, then as, Sosa correctly points out, it appears that Bonjour's argument that basic propositions require warrant of other propositions, rests on the claim that any justified belief is so in virtue of the non- epistemic fact of coherence, (í --> Jpí). If Bonjour's argument is directed towards the coherence theory, then it would on his view would require a reason or a justified belief that any belief cohering with other beliefs Is likely to be true, and this meta- belief would itself require a justified belief as a source of justification, Jp (í --> Jpí). Such a regress is vicious if the principle of supervenience is to be maintained by coherentism. This is a version of the doxastic ascent argument used against coherentism.(13) This argument shows that the doxastic ascent argument is uninteresting because it can be also used against coherentism, but it does not show that coherence cannot be used as a non-epistemic source of justification.(14)
IV. The Justification of Epistemic Principles
In
the previous section we have been discussing the structure of foundationalism
coherentism, now we turn to discuss how the foundationalist correctly justifies
and how the coherentist incorrectly justifies epistemic principles.
Epistemic principles for the foundationalism are second-level principles which justify first- basic propositions, for example "I have a headache" is a basic proposition at the non-epistemic level (belief) while "I am justified in believing that I have a headache" is at the first-level of epistemic propositions. But these propositions are not the ultimate epistemic principles of coherentism; they are just trivial propositions that refer to first-level basic propositions, they do not provide an explanation of why basic propositions of first-level foundationalism fall under them.
Foundationalists accuse coherentists of giving up supervenience, but how are epistemic principles justified within foundationalism? Coherentists accuse foundationalists of a circularity of their own. Coherentists claim epistemic principles must themselves be known (or justified) if knowledge (or justified belief) is to arise, and there is no way within the foundationalist framework for epistemic principles to be either justified or known.
In response the foundationalist(15) claims that because there are distinct levels of justification so that after first level basic propositions are known. Second-level foundationalism epistemic principles are justified and circularity is avoided; i.e., if a given principle is in fact true, then any belief that is a justified belief that is specified by the principles will be immediately justified. Following some revision of Van Cleve's position I hold that neither the principles nor claim that the belief satisfies the conditions down in the principle need to function as premises in the foundationalist argument for the principle.(16) An epistemic principle has the form "If...then P is justified for S". There are generation or immediate justification principles of the form "if... then P is justified for S" (If í then Jpí). In other words it says that the obtaining of whatever condition is specified in its antecedent is sufficient for P's beliefs justified for S. It is a logical truth that if X is sufficient for Y, then there is no other condition Z that is necessary for Y, unless Z is also necessary for X. Knowledge of a generation principle is not necessary for the obtaining of its non- epistemic antecedent. An example is "If I seem to have a headache, then the proposition 'I seem to have a headache' is justified for me". In other words if generation are true, then the obtaining of the condition mentioned in its antecedent P "I seem to have a headache" will be sufficient for P's being justified for S. But if this condition is sufficient, then no condition is necessary which is not also necessary for obtaining of antecedent. Hence, justified belief may arise in accordance with epistemic principles which are not themselves justified, they need only to be true.
Transmission or derivation principles are principles whereby justification is transmitted from some propositions to others. An example of a transmission principles is if the proposition is justified to me that l have a headache then the proposition that the proposition is immediately justified to me that l have a headache is justified for me" (If Jpí, then JpJpí). The antecedent of a transmission principle has to be true in order that the consequent is justified. The point of difference between generation and transmission principles is that the former has an non-epistemic antecedent while the latter has not, and for the former, the truth of the principle is sufficient while not so for the latter. Transmission principles are of the form in which terms of epistemic appraisal such as "evident", "certain" are in the antecedent of the principle. Transmission principles transmit justification from one principle to another. Foundationalists have both transmission and generation principles and coherentism must have both as well in order there: are non-epistemic facts upon which the coherence system of beliefs supervenes. The one generation principle of the coherence theory is of the form. if p c q then Jp (p ^ q), so that the system is based upon a non-epistemic fact of coherence.
Unlike the coherentist, the foundationalist, does not allow epistemic principles to function both as conclusions supported by singular perceptual beliefs and as support for those beliefs; i.e. the coherentist holds that beliefs about principles and singular beliefs stand in relation of mutual support. The circularity or "mutual support" stems from the fact that the coherentist justified inductively by claiming that a high proportion of such beliefs are true, but it is the epistemic principle of basic perceptual propositions that are reliable guides which justifies the basic perceptual propositions. Hence, the basic perceptual propositions are justified not because they are based on non-epistemic fact of sensory state of visual experience but upon knowledge the principle that perceptual propositions which contain those states, are reliable guides to truth. The foundationalist argument against the circularity of coherentism is that it is possible that the epistemic principles which a foundationalist defends could be sufficient to yield justification and, hence knowledge concerning matters of a specified sort, for example, perceptual knowledge, on the basis of such knowledge, one could determine how frequently perceptual beliefs of a specified sort turn but to be true. One could then note that perceptual propositions of the sort in question are the ones that are justified in terms of foundational , generation principles and if they are usually true, he generation principle would itself be justified as a guide to obtaining truth and avoiding error. As Van Cleve correctly indicates, the application of principle in such a manner is not circular. If the epistemic principles, are in fact, true, then they yield justified beliefs which simply results because certain propositions fall under the epistemic principles without hose principles conclude that having to be used explicitly as premises to the beliefs are justified; i.e., epistemic principles have a presumption of truth and confer positive epistemic value on the perceptual belief because he belief is formed when the agent does not have contrary evidence. An example would be a sober foundationalist who has a headache and could be justified in believing that he has a headache because he has a headache. If he gets drunk while he has a headache, he may after a while believe that he does not have a headache. He can form principles which justify his alternate belief that he does not have a headache. But those principles are false and his alternate belief is not justified because be has contrary evidence of the actual occurrence of the headache. But if he is a coherentist and he is drunk while he has a headache he could have an alternate belief system that he does not have a headache. His alternate belief system is justified even though he does not actually have a headache because the belief set follows from the coherence generation principle. But if a coherentist does admit to this, how does the actual fact of a headache fit in a coherentist belief system since a headache is neither a belief nor a relation amongst propositions? I contend that the headache does not fit in.
For coherentists like Lehrer and Cohen,(17) Van Cleve's epistemic principles in their view are inadequate because the principles do not provide explanation of the particular choice of antecedents. If the principles are true, it may be possible, after the fact to conclude that the propositions that fall under the principles arc more frequently true than false. If the truth of the propositions is guaranteed, then the question of why they are justified does not arise. However, once it is admitted that the foundational propositions are ones that carry no guarantee of truth, we must ask why those beliefs, which may be erroneous, have the sort of justification we require for knowledge. Moreover, Lehrer and Cohen claim that the explanation is not difficult to find, it is simply the high probability that such beliefs are reliable guides to truth.
I agree with their explanation but not with their demand that a theory of justification must be judged in terms of how well it explain why certain beliefs are justified and why some are not through a system of principles in which there is one ultimate principle. For Lehrer, it is justifiably believes P iff S takes belief in P to be part of a strategy for maximizing expected utility in truth-seeking. Coherentists find a related problem with foundationalism is that it amounts to sets of epistemic principles, one concerning perception, one concerning memory, etc. These principles tell us that beliefs are justified under certain conditions, but there is no general theory or single basic principle like Lehrer's that unites all the other principles. Coherentists have basic principles, for Lehrer and Cohen, it is avoiding error and accepting truth; these principles indicate the internal connection between justification and truth which i.e. what it takes for a belief to count as an instance of knowledge. The foundationalist response can be stated in two ways: (i) It is an epistemic fact that under certain conditions (stated in the antecedent of the principle) a person is justified (stated in the consequent) in believing basic propositions and no further explanation is required for no evidence justifies them, for example "If I seem to have a headaches then I seem to have a headache is justified for me"; (ii) Foundationalists do not intend to provide a one principle solution to the problem of the justification of epistemic principles. Donald Davidson in his infinite wisdom has aptly stated a similar position with respect to moral principles. I shall not argue the point here, but I do not believe any version of the single principle's solution, once its implications are understood, can be accepted: principles, or reasons for acting are irreducibly multiple.(18)
The foundationalist's response is that he does not have one true principle that unites all the particular principles because for him, those persons who are justified do not have to know the theory of justification in order that they are justified, and also even if the principles of justification are true, they do not explain the truth of basic propositions. The point is that epistemic principles are not ultimate principles for the foundationalist as they are for the coherentist because for the former basic propositions are ultimate and epistemic principles are not. Moreover, the explanation demanded by the coherentist of the justification of one ultimate epistemic principle cannot be met by the foundationalist's sets of principles because they do not function in a unified way.
V. Chisholm and Sellars on Frameworks
The
foundationalist can either justify his epistemic principles immediately in terms
a priori knowledge of necessary truths (if the analytic/synthetic or
necessary/contingent distinctions are valid) or mediately in terms of a
conceptual framework. For traditional foundationalism, immediate justification
in terms of a priori knowledge goes this way: necessary truths are known a
priori and they immediately justify epistemic principles though immediate
justification is not a criterion of necessary truth. Necessary truths are
analytic truths such as "All square are rectangular" or truths of logic such
tithe law of non-contradiction These truths are true in all possible worlds
including this actual world unlike basic propositions which are true only in
this world In the Logical Investigations Husserl in his critique of
psychologism validates a priori knowledge of epistemic or what he calls logical
principles by arguing that if the principles of logic are reduced to the
principle of thought, then principles of logic are relative to the human species
and are not universally true; i.e. the law of non-contradiction holds
universally and not just for humans. The problem with immediate justification in
terms of a priori knowledge is that what is known is a very small amount since
the conditions for memory, perception and other sources of knowing are
omitted.(19)
If there arc foundationalist epistemic principles, and if Quine is right with respect to the analytic/synthetic distinction and that a priori knowledge is otiose, then justification of epistemic principles has to be achieved in terms of our existing conceptual framework. I shall call this contemporary foundationalism. The contemporary foundationalist does not claim that lie knows his epistemic principles better than we know that this a sheet of paper or we have two hands, etc. Instead, he offers us an epistemological theory of what it is to be justified in believing that we have two hands or that this is a sheet of paper. This can be done by comparing rival theories of justification such as reliabilism, coherentism or pragmatism and showing that foundationalism is a more adequate theory of justification than its rivals. There may even be a better theory whim which has yet to be discovered.
A contemporary foundationalist epistemic theory is a hypothesis which is justified by our common sense beliefs. This seems circular that with our common sense beliefs to justify our epistemic theory which is supposed to justify our common sense beliefs. Is this a form of coherentism? Yes and no. It is a form of coherentism at the level of the justification of epistemic principles though the structure is that of formal foundationalism and it is substantive foundationalism at the level of basic propositions. Chisholm has provided a rational reconstruction of epistemology in which he presupposes that we have a framework of justified beliefs which we never made and his task is to provide the principles of justification Chisholm in "A Version of Foundationalism" has provided us with prima facie justified epistemic principles which at best silences the sceptic who doubts the existence of such principles. However, this does not silence the sceptic who denies the truth of the principles. But for Chisholm doing epistemology does not require the truth of the principles, which can be false, and moreover, the sceptic does not argue that we cannot have prima facie justified principles since they can be false. Epistemic principles are justified by fitting well into our conceptual framework which include beliefs about what we know and are justified to believe. With respect to the question of the soundness of the principles Chisholm's response is that truth and justification are separate. If conditions of justification arc conditions of truth, then our fully justified beliefs about the world cannot be false. Lehrer and Cohen agree, but Chisholm and I disagree by appealing to our epistemic boundedness and the denial of verificationism.
Sellars has taken a similar route to Chisholm in "More on Giveness: and Explanatory Coherence" though there are two major differences between them. Sellars claim that one must know the epistemic principles and the framework that justifies them in order that the principles are justified. This is Sellars' coherentism. Chisholm denies Sellars' claim. This is Chisholm's foundationalism. The other difference is that Sellars justifies the entire framework with the notion of effective agency. To be effective agents we need reliable cognitive maps of the environment in order we are effective. Chisholm justifies his framework in terms of common sense beliefs. Who has a better theory? I think Chisholm has for though he does not refute the sceptic with respect to knowledge, he does refute the sceptic in terms of the existence of epistemic principles. With respect to Sellars, the sceptic may charge that Sellars' justification of the entire framework which includes his epistemic principles fails, for the notion of effective agency is hard to accept. The sceptic claims that we are not effective agents since there is the danger of a nuclear holocaust from the malfunction of a computer or there is the increasing pollution of environment. The sceptic may even claim that though we may think that we are effective agents, in fact we have lost control of technology and so the notion of effective agency does not conform to the present image of the world. Sellars' response may be that the notion of effective agency is different from the application of the notion. The sceptic's response can be verificationist in that a concept does not have justification beyond our every day use of the concept which is just the application of the notion. Whether verificationism is right is the topic for another paper.(20)
Endnotes
1. L. Bonjour "The Coherence Theory of Empirical Knowledge", Philosophical Studies, 30 (1976), p. 2S5.
2. see footnote 11
3. B. Russell Human Knowledge (New York: 1948). p. 416 and C.I. Lewis, "The Given Element in Empirical Knowledge", Philosophical Review 61 (1952), p. 173. Lewis writes: "the probabilistic conception strikes me as supposing that it enough probabilities can be got to lean against one another they can all be made to stand up I suggest that, on the contrary, unless some of them can stand alone, they will all fall flat."
4. Hume's impressions of reflection are self-evident propositions but they are not empirical propositions. Nonetheless, they are derived from impressions of sensations.
5. cf. W. Alston, "Varieties of Privileged Access". American Philosophical Quarterly 8 (1971), p. 223-241 and R. Firth, "`Coherence, Certainty and Epistemic Priority", Journal of Philosophy 61 (1964), p. 545-546 and "Anatomy of Certainty", Philosophical Review 76 (1967), pp 3-27.
6. R. Chisholm. The Theory of Knowledge, (Englewoods, N. J.: Prentice Hall, 1966). p. 25. Chisholm's sample response in the second edition is nearly the same as the first: "What justifies me lit thinking l know that he has that disorder is the fact that it is evident to me that he has those symptoms". (p. 18).
7. R Firth, "Coherence and Epistemic Priority", p. 553.
8. Ibid., pp. 552-3.
9 .E. Sosa, "The Foundations of Foundationalism" Nous 14 (1980), pp. 54~550. Cf. Sosa's "The Raft and the Pyramid" in Midwest Studies in Philosophy. VoI. V. ed. French, Uehling and Wettstein, (Minneapolis: U. of Minnesota Press, 1980), pp. 3.26.
10. See E. Sosa's "Nature Unmirrored, Epistemology Naturalized", Synthese 55(1983), p. 49-72. Sosa claims that though Rorty denies that he is a substantive foundationalist, he is at is a formal foundationalist with respect to social justification.
11. In ethics, a utilitarian, for example, holds that the property of being good supervenes upon the property of producing more pleasure than pain. Other accounts of those properties upon which goodness supervenes are also possible, but it is one of the tasks of ethics is to discover what the properties are. In epistemology coherence is a "warrant-increasing" property just as producing more pleasure than pain is for the utilitarian because whatever ethical act has that as a property will be a good act on the utilitarian account. The meta-statement that "If S has coherence as a property then S is likely to be true" is a warrant- increasing property derivatively on the second level of justification because coherence is the basis property of the first level which justifies this statement, "If S has coherence, then S is justified". In the same way, for the utilitarian being an act of kindness is derivatively only a right making property on the second level of justification, since acts of kindness are grounded on ethical acts of the first level which in turn are grounded on the non-moral fact that more pleasure is better than pain.
Supervenience is, of course not peculiar to the coherence theory, the thesis of epistemic priority asserts that basic propositions have a warrant-increasing property which is independent of the relations and that they supervene on non-epistemic facts. Those who accept this view often hold justification supervenes upon one's present experience and one's beliefs about one's present experience. Yet some appear to hold that justification supervenes merely upon one's belief about one's present experience. Firth holds the latter view while Lewis holds the former view? That is, Firth holds to the view that we have beliefs about one's experiences while Lewis holds the view that we have experiences
12. Sosa, `"Raft and Pyramid", pp. 16-18.
13. Some philosopher: such as Sosa have called the doxastic ascent argument the intellectualist model of justification: that is, we make an explicit appeal to epistemic principles in order to acquire beliefs. For example, S must appeal to the epistemic principle in order for s to be justify in believing that he sees that it a tree from the belief that something looks like a tree, S must believe that the principle is included in his epistemic principles, and in order for S to be justified in holding the principles S has to appeal to other principles. The problem is that most people do not have beliefs about how they are appeared to as well as beliefs about their reliability as perceivers, when they have perceptual beliefs. The foundationalist thesis is that the rejection of the intellectualist model is also the rejection of any connection of justification and truth at the doxastic level.
14. Cf. N. Lemos. "Coherence and Epistemic Priority", Philosophical Studies, 41 (1982). pp. 299-313 for a similar conclusion.
15. James Van C1eve, "Foundtionalism, Epistemic Principles and Cartesian Circle", Philosophical Review, 88 (1979) pp. 5-91.
16. Ibid., pp. 75-78.
17. K. Lehrer & S. Cohen, "Justification, Truth and Coherence", Synthese, 53 (1983), pp. 203-205. See also S. Cohen. "Justification and Truth", Philosophical Studies 46 (1984), pp. 279-295
18. D. Davidson. "How is Weakness of the Will Possible" in Actions & Events (Oxford: Clarendon Press, 1980). p. 34.
19. R. Chisholm. The Foundations of Knowing (Minnesota: University of Minnesota Press, 1981). pp. 187-188.
20.
1 would like to thank Steven De Haven, University of Alberta, Bruce Hunter,
University of Alberta and Karl Pfeifer, University of Saskatchewan for their
comments on earlier drafts of this paper.