On the Problem of Induction
A Critique of Nelson Goodman’s Concept of the New Riddle of Induction
The development of the method of induction has been privy to the presentation and solution of riddles. At the initial level of its development, it has been privy to the old riddle of induction discovered by Hume. After the solution of the former riddle, however, a new riddle of induction was discovered by Nelson Goodman.
In lieu of this, this paper opts to consider the development of the method of induction as a methodology defined by Hume and Goodman’s conception of the Inductive method.
Induction refers to “a method of reasoning by which a general law or principle is inferred from observed particular instances” (Flew 171). The method of inductive inference may be considered as the primary means through which justifications are formulated to show the relationship of evidence towards particular assumptions (Norton 2). The process of induction, in this sense, may be seen to arise whenever we note that evidence lends support to a hypothesis while in the process failing to establish its deductive certainty. It was such a formulation of the method of induction that enabled the conception of the first riddle. What follows is a presentation of the main arguments of the aforementioned riddle as formulated by David Hume.
Hume argued that since no necessary connections exists between empirical phenomena, it is always possible that a future observation will prove our inferences wrong no matter how appealing it may have been or how richly supported by past observations. This problem, in the more recent formulations of the problem has been referred to as the uniformity principle [in this sense the lack of such uniformity]. According to the argument, nature has no uniformity. If such is the case it thereby follows that there is no voucher that which ensure the consistency of man’s most refined predictions. It might be argued that such an assumption has never been denied in the formulation of predictions however there has been agreement regarding the results of such an agreement [or lack thereof] within the province of induction.
To some, it means that induction is never valid or justified, while to others, it means that induction simply calls for different standards of validity (Landesman 164). The latter view strips the aforementioned riddle [Humean riddle] of its problematic context. This is evident if one considers that since the rules of deductive validity are inapplicable to induction, it cannot be a problem that inductive inference is unavoidably attended by the possibility that a future observation may prove it wrong (Goodman 4). The old riddle is then dismissed because it cannot possibly be the genuine problem of induction.
Fact, Fiction, and Forecast present Goodman’s construal of what he refers to as the new riddle of induction. After refuting the old riddle of induction [the refutation of which is evident in the former paragraph], Goodman proceeds to outline what he takes to be the genuine problem of induction and its tentative solution. The problem of induction, he writes, is a problem of demonstrating the difference between valid and invalid predictions (Goodman 4). According to Goodman, a prediction is valid if it conforms to a valid rule of induction, and a rule is valid if it yields valid predictions.
He acknowledges that such an assumption is characterized by circularity however he notes that it is important to perceive such a conception of the problem in terms of the conceptions of justifications for arguments. Goodman notes that inductive predictions based on past regularities work better than those based on any other alternative. If such is the case, the rules for formulating predictions must be constructed in such a way that they will coincide with common practices of inductive reasoning.
This, on the other hand, is further developed by the quality of predictions, which it produces. This is clearly explicated by Rubenstein as he notes, “the centerpiece of a valid inductive logic [according to Goodman] is its reliance on past regularities, and the prescriptive mandate of inductive validity is inseparable from a descriptive account of how inductive judgments are commonly made” (39). This has been the result of Goodman’s dissolution of the old riddle of induction. What follows this is Goodman’s explication that the most promising solution of the aforementioned riddle is untenable. It is through the introduction of such untenability that Goodman presents what he perceives to be the new riddle of induction.
Goodman presents two hypotheses that are to be addressed through the use of the inductive method. One says that all emeralds are green and the other says that all emeralds are grue, where grue is said to apply to all things examined before t just in case they are green but to other things just in case they are blue (Goodman 10). Both hypotheses seem to be equally well supported by the evidence: all emeralds examined prior to t have been found to be green and grue. However, the two hypotheses are mutually exclusive. If emeralds are grue, they will be blue at t and thereafter, but if the alternative hypothesis is correct, they will be green. Thus, we are left with the paradox that Goodman christened the ‘new riddle of induction’.
We cannot, after all, justify induction by appealing to past regularities. However, the reason, according to Goodman, is not the lack of the elusive uniformity principle, but the previously unrecognized ubiquity of regularities. According to Goodman, regularities exist where one finds them. In relation to this Goodman states that one, however, finds them everywhere (12). If such is the case, it therefore follows that it is useless to base inductive validity on past regularities since it is not possible to predict and hence distinguish which regularities are valid and invalid.
At this point, I would like to present a summary of the aforementioned discussion. In the aforementioned discussion, Goodman believes that the old riddle [the Humean riddle/the uniformity principle] has been dissolved and that induction is justified by past regularities. The only remaining difficulty he sees, however, lies in finding a rule for distinguishing between regularities that do and do not yield valid inductive predictions. As was noted in the above discussion, the possibility of such is not possible. This is evident if one considers that regularity necessitates the occurrence of acts of inductive inference. Therefore, the genuine problem of induction cannot be the distinction between the distinction of regularities that do or do not yield valid inductive predictions since the specification of such necessitates the formulation of inductive inferences.
As I reckon, Goodman aforementioned conception fails to account for the process of induction. It is important to note that Goodman contends that induction begins with regularity. Rubenstein notes, “induction does not begin with regularity – it ends with it” (44). The failure to consider this leads Goodman to misconstrue the problem of induction. It is important to note that experience of reality does not necessarily start with regularities but rather with individual observations. The role of induction, in this sense lies in providing us with justified methods that allows us to posit the observations that we will account for as regularities. Goodman, however, failed to account for this.
In addition to this, it is important to note that such a failure can also be traced to Goodman’s assumptions regarding the process in which individuals formulate inferences. Goodman’s error is compounded when he makes a distinction between identifying regularity and projecting it. Once we have decided that our observations represent regularity, it is automatically projected in both temporal directions. This is, in fact, what we mean by applying the term regularity to our data.
Furthermore, Stich and Nisbett contend that the “equilibrium with inductive practices” that Goodman posited as a necessary aspect in formulating a valid inductive methodology is “neither necessary nor sufficient for a rule of inductive inference to be justified” (194). They argue that such an assumption fails to consider that “human subjects regularly and systematically make invalid inferences” and that there an instance wherein human reasoning enables an individual to “accept invalid rules and reject valid one’s that ought to govern the inference at hand” (Stitch and Nisbett 194).
In summary, the aforementioned paper presented Goodman’s arguments in relation to his conception of the new riddle in induction. Such a riddle, however, under scrutiny may be seen as based upon a mistaken assumption of the justification process of beliefs that necessitates the introduction of information garnered through the method of induction. This is evident, for example, if one considers the manner in which observations enable the formulation of regularities and not the other way around. An analysis of Goodman’s supposed riddle of induction thereby leaves the reader wondering if such a riddle may be considered as a valid concern for the adherents of the inductive methodology.
Flew, Anthony. A Dictionary of Philosophy. London: Pan Books, 1983.
Goodman, Nelson. Fact, Fiction, and Forecast. Massachussets: Harvard University Press, 1983.
Landesman, Charles. Skepticism: The Central Issues. London: Blackwell Publishing, 2002.
Rubenstein, Arthur. “Induction, Grue Emeralds and Lady Macbeth’s Fallacy.” The Philosophical Quarterly 48.190 (Jan. 1998): 37-49.
Stitch, Stephen and Richard Nisbett. “Justification and the Psychology of Human Reasoning.’ Philosophy of Science 47.2 (Jun. 1980): 188-202.