George Steven Botterill

The thesis of underdetermination presents a major obstacle to the epistemological claims of scientific realism. That thesis is regularly assumed in the philosophy of science, but is puzzlingly at odds with the actual history of science, in which empirically adequate theories are thin on the ground. We propose to advance a case for scientific realism which concentrates on the process of scientific reasoning rather than its theoretical products. Developing an account of causal–explanatory inference will make it easier to resist the thesis of underdetermination. For, if we are not restricted to inference to the best explanation only at the level of major theories, we will be able to acknowledge that there is a structure in data sets which imposes serious constraints on possible theoretical alternatives. We describe how Differential Inference, a form of inference based on contrastive explanation, can be used in order to generate causal hypotheses. We then go on to consider how experimental manipulation of differences can be used to achieve Difference Closure, thereby confirming claims of causal efficacy and also eliminating possible confounds. The model of Differential Inference outlined here shows at least one way in which it is possible to ‘reason from the phenomena’.

There are two epistemological problems connected with dreaming, which are of different kinds and require different treatment. The internal problem is best seen as a problem of rational consistency, of how we can maintain all of: Dreams are experiences we have during sleep. Dream-experiences are sufficiently similar to waking experiences for the subject to be able to mistake them for waking experiences. We can tell that we are awake. (1)-(3) threaten to violate a requirement on discrimination: that we can only tell Xs from Ys if there is some detectable difference between Xs and Ys. Attempts to solve the problem by Descartes and Williams are considered. It is suggested that if we take account of levels of epistemic risk, we can use Descartes's criterion of lack of coherence, at least with hindsight - which is the time when we need to use it.

Davidson argued that the fact we can have a reason for acting, and yet not be the reason why we act, requires explanation of action in terms of the agent's reasons to be causal. The present paper agrees with Dickenson (_Pacific Philosophical Quarterly_, 2007) in taking this argument to be an inference to the best explanation. However, its target phenomenon is the very existence of a case in which an agent has more than one reason, but acts exclusively becaue of one reason. Folk psychology appears to allow for this phenomenon. However, appreciation of 'rationalization' as a form of contrastive explanation reveals the existence of the Davidsonian possibility to the problematic. Claims that 'I did it because of R 1, not because of R 2 ' are entertained in folk psychology, and may be sincere or insincere. But as reports of conscious practical reasoning, even when sincere, they are not authoritative about the mechanism of motivation.

To give a causal explanation is to give information about causal history. But a vast amount of causal history lies behind anything that happens, far too much to be included in any intelligible explanation. This is the Problem of Limitation for explanatory information. To cope with this problem, explanations must select for what is relevant to and adequate for answering particular inquiries. In the present paper this idea is used in order to distinguish two kinds of causal explanation, on the grounds of systematic differences in their conditions of relevance and adequacy. It is further argued that these two forms of causal explanation are interdependent and their interaction provides an instrument through which causal knowledge is acquired and enhanced. What we understand causation in the world to be is neither unconditioned regularity, nor counterfactual dependence, but the sum of correct answers to explanatory inquiries of these two interdependent kinds.