I’m continuing to read James F. Harris’s Against Relativism. One of the interesting things about this book is that Harris looks at several relativistic philosophers and critiques them in various chapters. One such critique comes against Nelson Goodman in Chapter 3 of Against Relativism.
Goodman’s relativism is based on an attack on induction; one that is a result of the thoughts of David Hume. Induction, for those who are not aware, is basically how science works, by infering from the observed to the unobserved. (For more, check out this wiki article.)
The problem with induction is that one must be able to justify one’s inductions. As Harris explains:
The difficulty is one of supplying a rationale or justification for the claim that reports of observations concerning known or observed cases provide some logical basis for making judgments about unobserved cases. Induction is the process of inference whereby one concludes that unobserved cases will resemble observed cases in some crucial respect. But, as Hume’s question goes, why would one ever be justified in believing this to be so? The question is neither simply why one believes that it is so nor simply why one trusts in or uses induction, but rather why such beliefs are epistemologically justified.
Harris, James F. (1992).  Against Relativism: A Philosophical Defense of Method, LaSalle: Open Court (pp. 51-52), emphasis in original
One example of this problem is found in a quote of Charles Sanders Peirce in the book:
A chemist notices a surprising phenomenon. Now if he has a high admiration of Mill’s Logic, as many chemists have, he will remember that Mill tells him that he must work on the principle that, under precisely the same circumstances, like phenomena are produced. Why does he then not note that this phenomena was produced on such a day of the week the planets presenting a certain configuration, his daughter having on a blue dress, the milkman being late that morning and so on? (6.413)
(ibid, 62)
In short, induction is not a clear-cut thing. And this brings up Goodman’s “Grue Paradox.” The Grue Paradox can be framed in the following manner.
First, we have the usual predicate: All emeralds that have been observed before a specific time t (ie. “now”) are green. Therefore, we generalize: “All emeralds are green.”
However, Goodman stipulated a second predicate: “All emeralds observed before a specific time t are green, but after t they will be blue.” Thus, we generalize that “All emeralds are grue.”
As Harris states:
Exactly the same evidence which confrims ‘all emeralds are green’ also confirms ‘all emeralds are grue’. That is, for each observation of an emerald which is described by the true observation ‘a is green’, ‘b is green’, and so forth, there is a corresponding true obeservation statement ‘a is grue’, ‘b is grue,’ and so on. Consequently, the two generalizations, ‘all emeralds are green’ and ‘all emeralds are grue’ are each equally confirmed by any and all of exactly the same observations (before t).
Having two different hypotheses confirmed by exactly the same data may not be troubling in some circumstances; however, in the present case the situation becomes particularly intolerable given the mutual inconsistency of the two hypotheses. A green emerald will be green after t, but a grue emerald will be blue after t given the definition of ‘grue’. Whichever of the two hypotheses we might regard as confirmed, the other is equally confirmed, and hence, we are left in the very unenviable position of being unable to give an evidential reason for choosing one hypothesis and rejecting the other since, to emphasize the point, there is absolutely no evidence which confirms one which does not also confirm the other.
(Ibid, 55)
This highlights the problem with induction quite nicely. Harris, following Peirce, tries to solve this by appealing to abduction. In this case, however, it appears that Harris surrenders the ability to affirm the truth-detecting ability of science. That is, in separating abduction from induction, induction loses its ability to actually declare which hypothesis ought to be considered and which ought not be.
This is a problem for Harris because he writes: “Briefly, I will suggest a way of understanding Goodman’s new riddle of induction…such that none of the threatening relativistic consequences for science follow” (ibid 59-60). If Harris’s solution does not remove the “threatening relativistic consequences” then either relativism is true (which I, personally, disagree with of course) or else Harris’s solution has a problem.
So what is his solution?
It appears that Goodman’s new riddle of induction results from…the “failure to distinguish the essentially different characters of different elements of scientific reasoning.” The new riddle of induction–whether it is a riddle at all or not–is not, in any case, a riddle of induction. Goodman obviously treats abudction and induction as “a single argument” and conflates abduction and induction. If we adopt Peirce’s distinction, then the new riddle of induction is properly viewed as a riddle of abduction, a meta-scientific riddle. …For Hume’s problem to arise, the hypothesis must already be in place. Goodman’s new riddle of induction is clearly a problem concerning how a hypothesis gets selected for confirmation. Will we select ‘all emeralds are green’ or ‘all emeralds are grue’? This question is an issue of abduction, not induction.
(ibid, 60-61)
Everything Harris said here is accurate. Goodman’s argument does not deal so much with the method of induction itself as it does with which hypotheses should we use? This is, indeed, a meta-scientific question. As Harris continues:
Goodman tells us that is is because ‘all emeralds are green’ and ‘all emeralds are grue’ are equally “confirmed by evidence statements describing the same observations.” now how is it that the two competing hypotheses get confirmed in order to generate the paradox unless induction is being used with each hypothesis? In other words, the new riddle of induction becomes a riddle only if induction itself is used since both hypotheses must be confirmed (by induction) in order for Goodman’s paradox to arise. This is clear evidence that the problem raised by Goodman’s new riddle of induction is on the meta-scientific level involving the selection of hypotheses rather than on the level of confirmation of hypotheses by induction.
(ibid 61)
Again, Harris is correct. And in being correct, he seals off the ability for science (which he equates with induction) to determine which hypothesis is correct in a given system. In other words, the results of induction are relative to the hypthesis that one begins with.
This relativism is one that Harris acknowledges:
So far as raising serious questions of relativism for induction is concerned, the new riddle of induction is a sheep in wolf’s clothing because the alleged pernicious, relativistic consequences of the grue paradox for induction do not follow. The scientific method which uses induction in the process of confirmation for providing epistemological warrant for a hypothesis must be presupposed by the grue paradox since, in order for the paradox to be generated, conflicting projections must be equally confirmed. Seen in this manner, the new problem of induction is neither new nor a problem of induction. It is the old problem of theory formation, Peirce’s process of abduction, and induction is not threatened by new charges of relativism. Surely, induction is relative to a particular theory, but this simply amounts to saying that confirmation is relative to a particular hypothesis.
(ibid, 62)
Under these circumstances, however, I fail to see how the charge of relativism is defused. It is moved to a different position to be sure; but it remains there. Now if all that Harris wishes to accomplish is to say that induction as a method itself is not relative, then that is fine…but with the above he is acknowledging that induction as a method is insufficient to give us any actual knowledge. That is, it is the hypotheses that are chosen and then put into the inductive method that determine what the results would be. And that seems, at least based on the summary view that Harris provides of Goodman (I confess I have not read Goodman myself yet), that this is exactly what Goodman was arguing in the first place.
In any case, I do wish there were more folks like Harris who would acknowledge the meta-scientific aspect to scientific theory. If there is one thing the discussion regarding Goodman can teach us, it’s that scientific theory (induction) alone cannot provide any knowledge and that science is only as good as the controlling philosophy it has. After all, induction can prove two sides of a contradiction if it is thought to be an actual epistemological end point.
