TWO CAUSAL MODELS

The final two models I shall discuss seem to offer a solution. They satisfy NES, violate the a priori requirement, and yet avoid the previous prob­lems. However, whether they provide accounts that would be welcome to most of those seeking models of explanation is quite dubious.

a. Brody’s Causal Model

As in the case of his essential property model, Brody regards this as pro­viding a set of sufficient conditions for explanations of particular events. These conditions are those of the basic D-N model together with the

Causal condition: The explanans “contains essentially a description of the event which is the cause of the event described in the explanandum.”^[82]

To see how this is supposed to work let us reconsider

(1 ) Jones ate a pound of arsenic at time t.

Anyone who eats a pound of arsenic dies within 24 hours.

Therefore,

Jones died within 24 hours of t.

The problem we noted with the basic D-N model occurs, e.g., if both premises of (1) are true but Jones died within 24 hours of t for some unrelated reason. Brody's causal condition saves the day since in such a case the event described in the first premise of the explanans was not the cause of the event described in the explanandum. Hence, on this model we cannot conclude that the explanans if true correctly explains the explanandum.

The present model, like those in section 5, violates the a priori require­ment. Whether the explanans if true correctly explains the explanandum is not an a priori question, since the causal condition must be satisfied; and whether the explanans-event caused the explanandum-event is, in general, an empirical matter. It is not completely clear whether Brody wants to exclude from the explanans itself singular causal sentences that entail the explanandum, but I shall consider that version of his model which makes this exclusion.

Woodward is another modelist who proposes the need for a causal condition to supplement the basic D-N model and his own functional interdependence condition:

These examples suggest that a fully acceptable model of scientific explana­tion will need to embody some characteristically causal notions (e.g., some notion of causal priority), or some more generalized analogue of these (e.g., some notion of explanatory priority). (p. 53)

However, unlike Brody, he leaves open the question of how such a causal condition should be formulated.^[83]

b. The Causal-Motivational Model

Here, as in the D-N motivational model (section 3), the explanandum is a sentence saying that some agent acted in a certain way.^[84] The explanans con­tains a sentence attributing a desire (motive, etc.) to that agent, and a sen­tence attributing the belief to that agent that performing the act described in the explanandum is, in the circumstances, a (the best, the only) way to satisfy that desire. Thus the explanandum might be a sentence of the form

Agent X performed act A,

and the explanans might contain sentences of the form

X desired G.

X believed that doing A is, in the circumstances, a (the best, the only) way to obtain G.

Unlike the D-N motivational model, however, a law in the explanans relating beliefs and desires to actions is not required. What is required is the satisfaction of a

Causal condition: X’s desire and his belief (described in the explanans) caused X to perform act A.

The counterexample cited earlier against the D-N motivational model is now avoided, since in that example it was not the agent's belief and desire mentioned in the explanans, but some other belief and desire, that caused him to act. As in the case of Brody's causal model, the a priori require­ment is not satisfied, but NES is.

I shall not here try to defend or criticize these two models.^[85] I shall assume for the sake of the argument that each avoids, or can be modified so as to avoid, the kind of problem I have been concerned with.

(2) Jones' eating a pound of arsenic at time t caused him to die within 24 hours of t.

(3) Smith's desire to buy eggs and his belief that going to the store is the only way to do so caused him to go to the store.

But these are singular explanation-sentences that entail the explanandum. To be sure, neither model requires such sentences to be in the explanans. Still, in each model to determine whether the explanans if true correctly explains the explanandum one has to determine the truth of such sen­tences. I am not criticizing the models on these grounds. But I believe that many of those who seek models of explanation will want to do so. They will say that in order to know whether the explanans in such a model cor­rectly explains the explanandum one has to determine, independently of the truth of the explanans, the truth of sentences of a sort these modelists want to exclude from the explanans itself. Morever, they will point out that there is not much difference between determining the truth of (2) and (3), on the one hand, and that of

Jones' eating a pound of arsenic at t explains why he died within

24 hours of t (or Jones died within 24 hours of t because he ate a pound of arsenic at t)

Smith's desire to buy eggs and his belief that going to the store is the only way to do so explains why he went to the store (or, Smith went to the store because he had this desire and belief ) on the other. Models which impose these causal conditions, they are likely to say, provide insufficient philosophical clarification for the concept of explanation, even though the explanation-sentence expressing the causal relationship is not itself a part of the explanans.^[86] If a central aim of modelists is to define terms such as ‘explain’, ‘because', and ‘causes', this excludes their employment as primitive notions in the explanans or, as in the present case, in the conditions of the model.

Regardless of whether we view such criticism as important, the present models are of interest because to avoid the sorts of problems raised in sec­tion 4 these models, unlike their D-N ancestors, require establishing the truth of empirical sentences to determine whether the explanans if true correctly explains the explanandum. In this respect they are like Salmon’s S-R model and Brody’s essential property model. However, unlike the latter, the empirical sentences whose truth they require establishing are themselves singular sentences that entail the explanandum.

7.