Some philosophers, such as Sellars, Quine, and Harman, for example, have appealed
to the simplicity of the overall system to supply the needed additional ingredients 17 Of two systems both of which have a maximum of explanatory coherence, the simpler of the two is the one providing complete justification for beliefs within it. There are some objections to this strategy. First, simplicity is both obscure and complex. The complexity of simplicity results from the different ways in which one System can be simpler than another and from a certain stress between these modes of simplicity.
One system may be simpler than another in terms of the postulates of the system; in terms of the basic concepts of the system; or, in terms of the ontology of the system.
We have, at least, postulational, conceptual, and ontological simplicity to consider, and these modes of simplicity may conflict. We sometimes purchase conceptual simplicity at the cost of multiplying entities in our ontology. Moreover, the notion of simplicity is hardly pellucid. It is
-108-difficult even on intuitive grounds, to judge when one system is simpler than another.
When a philosopher says his system is simpler than another, one may fairly suspect him of special pleading for the sort of system he prefers. Perhaps there is some
common feature of such preferred systems. Or maybe such preferences are shaped by the cognitive fashions of the decade. No matter, we may reasonably doubt whether there is any sufficiently articulate conception of simplicity to which impartial appeal could be made in choosing between explanatory systems.
Even if we were to grant, however, that there is some serviceable conception of
simplicity, this would fail to resolve the problem before us. There may be two systems that are not only maximal with respect to explanatory coherence but are also minimal with respect to complexity. If we have two systems that are equally coherent and equally simple, we shall have no way of deciding which system provides complete justification for the beliefs within it. Moreover, we actually complicate matters by introducing the concept of simplicity. Now we must balance simplicity against
coherence when, for example, one system is slightly more coherent and leaves more unexplained, while the other is slightly simpler and presupposes a smaller ontology.
Finally, the appeal to simplicity exacerbates a problem we left unsolved above, namely, that we may justify beliefs by depleting a system of statements to be
explained. By rejecting concepts and entities, we can obtain a simpler system as well as a more coherent one. If we seek both simplicity and coherence, we shall have the very strongest motive for rejecting observation statements for the purpose of reducing what needs to be explained, thereby obtaining greater explanatory coherence and simplicity. We again confront the sterile simplicity of a system confined to one theory, one law, and one set of confirming singular statements. Everything else may be
hygienically disposed of, to avoid explanatory untidiness, and thus keep the system clean and neat.
The authors cited appeal to a principle of conservation in an effort to escape the unwanted diminishment of the system. Sellars stresses the need to conserve
observation statements. 18 Quine and Harman refer to a principle of conservativeness or laziness in the general retention of beliefs. 19 If we apply their remarks to the problem before us, it is proposed that if two systems are equal in explanatory
coherence and simplicity, and all others are less coherent and less simple, then that
system provides complete justification for beliefs within it which conserves what we believe, at least among statements of a specified variety.
The primary problem with this proposal is simply that it is a principle of-epistemic conservatism, a precept to conserve accepted opinion. Sometimes, such a precept may provide good counsel, but often it will not. The overthrow of accepted opinion and the dictates of commonsense
-109-are often essential to epistemic advance. Moreover, an epistemic adventurer may arrive at beliefs that are not only new and revelatory but also better justified than those more comfortably held by others. The principle of the conservation of accepted opinion is a roadblock to inquiry and, consequently, it must be removed.
The preceding remarks are less than argument. Moreover, this principle of
conservation, though wide of the mark, embodies at least one important insight, to wit, that whether a person is completely justified in accepting something depends on her actual beliefs. Indeed, the fact of belief itself, the subjective reality of conviction, provides the basis for a satisfactory coherence theory of justification. Such a
coherence theory of justification based on the existence of belief and upon the comparative degree of such beliefs will find maxims of conservation and stability unnecessary and unwarranted. On the contrary, such a theory contains within it an explication of the way in which shifts and changes of belief, however radical, bring with them changes in what a person is completely justified in accepting. Before turning to the development of these ideas, however, it will be useful to have a summary of the results of this chapter before us.
Summary
We have found three major reasons for rejecting the explanatory coherence theory of justification developed at the beginning of the chapter. First, the explanatory
coherence of a system could be increased by decreasing what needs explanation. We thus reduce the problem of explanation by systematically denying the truth of those statements describing whatever is unexplained until we obtain a very simple system in which everything is perfectly explained because there is almost nothing to explain. No explanatory function or role of statements suffices to prevent this artificial
manipulation of explanatory systems. Second, we found examples of statements and beliefs that were completely justified by general statements within a system, such as the conclusion derived from the Pythagorean theorem, quite independently of any explanatory role or function of such statements. Finally, systems may tie for the award of being the system with a maximum of explanatory coherence. A statement
completely justified with respect to one such system is not justified with respect to another. Indeed, some statement inconsistent with the first may be completely justified in another equally maximal system.
All these difficulties spring from the same source. Having abandoned a foundation theory in which justification is built upon self-justified basic beliefs, we are led by the explanatory coherence theory to build