Eliminative inferences and the problem of unconceived alternatives

Abstract

The argument of the inference to the best explanation states that (i) because an evidence (E) needs explanation and (ii) because a hypothesis (H), based on true background knowledge, offered a better explanation for (E) than its rivals, it follows that (iii) we have good reason to believe that H is true. This argument is presented for the purpose of giving reasons in favor of the reliability of scientific methodology: the best theory would have emerged in a struggle context in which rival theories were eliminated. However, to antirealists this argument has serious flaws. One problem with this argument is the assumption (ii) which describes a contest between rival hypotheses; according to Kyle Stanford, the historical record shows that sometimes the struggle doesn’t occur, because some rival hypotheses to a hypothesis proposed were not conceived, and therefore the reliability of science would not be, at least in some cases, a consequence of eliminative inferences; therefore we would be facing the “problem of the unconceived alternatives”. In this article we aim to describe briefly the argument of the inference to the best explanation. Then we present the problem of unconceived alternatives. After that, we show a case study provided by Stanford to support his position. Finally, in concluding remarks, we point out a little problem in the Stanford’s proposal.

Keywords: inference to the best explanation, scientific realism, unconceived alternatives.