RELATIVIZATIONS

fare, and I will suggest strengthening it in a moment. My A-concepts of evidence yield different results depending on which concept is chosen. Veridical evidence A', which requires the truth of h, classifies h as a speculation for “theo- rizer” P whenever h is false, even if P knows facts e that con­stitute evidence that h in the other A-senses, and P is justified in believing that e also constitutes veridical evidence that h.

For example, Isaac Newton believed that the fact that (e) the planets lie on the same plane and rotate around the sun in the same direction is evidence that (h) God exists and designed it that way.^[24] (He claimed this because, he said, this can't be due to “mechanical causes, since comets go freely into very eccentric orbits and into all parts of the heavens.”) Newton would have denied vehemently that he is speculating in this case, since he regarded h as “deduced from the phenomena.” Many would reject Newton's claim that e is evidence that h, and say that when Newton introduces h to explain e, he is indeed speculating. When Newton claims that e is evidence that h he should be understood using the concept of subjec­tive evidence. And his denial that he is speculating should be understood using the same concept of evidence in (Spec). In the subjective sense (one reflecting what he believed he was doing or intended to be doing), Newton was not speculating.

Newton's critics who claim that he is speculating should be understood as using either the concept of ES-evidence or potential evidence in (Spec).^[25] They are saying that in one or the other of these senses of evidence, Newton was indeed speculating (whether or not he believed he was).

Now, two questions need to be addressed: (1) Using the Bayesian definition (B), how much increase in probability must e give to h so that P's knowing that e is B-evidence that h prevents h from being classified as a speculation for P? Alternatively, using my definitions supplied in the (As), how probable must it be that there is an explanatory con­nection between h and e, given e, so that knowing that e is A-evidence that h prevents h from being classified as a spec­ulation for P? (2) What sorts of considerations will increase the probability of an assumption h? Alternatively, what sorts of considerations will make it such that there is a high prob­ability that there is an explanatory connection between h and e, given e?

I will not attempt to give question (1) a detailed answer. We could talk about degrees of speculation, depending on how high the probabilities are. But let's simplify the discus­sion, make things more precise, and set the bar reasonably high by requiring that the probability involved be greater than %. If we can suppose, as I do, that evidence must pro­vide a good reason for believing a hypothesis, and that this requires that the evidence makes the hypothesis at least more probable than not, this will set a minimal standard for what is and is not a speculation. Assumption h is a speculation for you if you introduce h, under “theorizing” conditions, without knowing that there is evidence that provides a good reason for believing h. Using the Bayesian idea, we upgrade (Spec) and say this: h is a speculation for person P, who introduces h (under “theorizing” conditions), if and only if P does not know that there is some fact that increases h’s prob­ability so that the latter is greater than %—i.e., more likely than not.

It is question (2) that I regard as the central one. In the following section, l will look at various answers proposed to question (2). These answers are based on the use of an objec­tive, rather than a subjective, concept of evidence, and hence of speculation. They propose ways of obtaining evidence that are supposed to hold no matter what particular individuals believe about what is evidence for what. This is not at all to reject the idea that there are subjective concepts of evidence (e.g., my subjective A-concept, or the Bayesian one under­stood subjectively). Nor is it to reject the idea that there are subjective concepts of speculation (e.g., ones that accord with (Spec) in which evidence is understood in terms of my subjective A-concept or in terms of a subjective Bayesian one). My claim is that there are objective ones as well, and that these are the most important and interesting ones, espe­cially in understanding the controversies noted in section 3 about the legitimacy of speculating and how speculations are to be evaluated. Later, on the basis of the answers proposed for question (2) I will argue that if an objective B-concept is employed in understanding what it is to be a speculation, it will need to be augmented by incorporating an explanatory idea present in the A-concepts. To simplify the discussion, the objective A-concept I will focus on is potential evidence, since the others are defined using it. The question now is how to get objective evidence in one of these senses.

8.