Possibility Shifting, Error Slicing, and Rendering Relevant: On Moral Encroachment, Corroborating Evidence and a New Kind of Testimonial Injustice

Georgi Gardiner (University of Tennessee, Knoxville, Oxford University)

I first sketch the relevant alternatives framework and contrast it with the more familiar “quantifiable balance” conception of epistemic support. The relevant alternatives framework holds that claim p is established to an epistemic standard, L, only if the evidence eliminates the L-relevant error possibilities. On the “quantifiable balance” model, epistemic support is a matter of probability given the evidence, and epistemic standards—such as legal standards of proof—correspond to numerical evidential probability thresholds.

I explain two advantages of the relevant alternatives framework. Firstly, the quantifiable balance model is unable to capture the important distinction between ‘threshold-shifting’ and ‘alternative-shifting’ effects. Pragmatic encroachment holds that when error costs are high, more evidence is required to know. If the demand is simply for more evidence, without specification of which particular error possibilities are newly relevant, then this exemplifies threshold-shifting. But now consider moral encroachment, the view that moral features can render some beliefs unjustified. I argue that this mechanism cannot be understood as an increase in overall threshold. Instead, particular error possibilities are rendered relevant by moral factors. Regardless of whether one endorses these encroachment views, the distinction is important.

Secondly, the relevant alternatives framework better explains the epistemic significance of corroborating evidence. The value of secondary evidence is not fully captured by an increase in probability. I close by articulating and briefly addressing a question, which I call the ‘constructivist question’: If an error possibility is often treated as relevant or true, can this by itself render the error possibility relevant? If many people express concern that rape accusers lie for financial gain, for example, does the error possibility thereby become relevant? Answering ‘no’ endorses epistemic hubris and dogmatically ignoring evidence. Answering ‘yes’ suggests a novel kind of testimonial injustice.