MBBS Curriculum 2020

Induction and Necessary Connections in Medical Research

21 January 2020 - 17:30-19:00

Lecture: Marius Backmann, London School of Economics

Bush House (S) 2.02, Strand Campus

If you do not have a KCL ID, please register (free) at this Link.

(See this event also on Facebook)

 
Description:

Some necessitarians have claimed that they could justify induction by
introducing necessary connections. By analysing the reasoning in randomised
clinical trials (RCTs), I argue that this view does not accurately represent
scientific practice.

The basic model of necessitarian solutions to the problem of induction is as
follows: First we infer from the fact that all Fs have so far been Gs via an
inference to the best explanation (IBE) that there is a necessary connection
between F-ness and G-ness. We then deductively infer from this necessary
connection that all Fs are Gs.

Nancy Cartwright and Eileen Munro offer an idealised reconstruction of
randomized clinical trials broadly along these lines. First, we infer from
evidence that a treatment has a ‘stable capacity’, i.e. a modal dispositional
property, to produce an outcome. Second, we deductively infer the efficacy of
the treatment outside the test environment from the existence of this stable
capacity. Cartwright and Munro argue that RCTs alone are no basis to support
these sorts of inferences, and hence do not deserve the status of a gold standard
for medical research.

Against this, I argue we should not try to give a deductive reconstruction of
RCTs. We ampliatively infer the causal relevance of the treatment in the sample
from the fact that the desired outcome is more prevalent in the test group than
in the control group. The further inference that the treatment will be causally
relevant in the population will also always be ampliative, because we cannot
possibly have the necessary information to make it deductive.
Moreover, the necessitarian analysis of inductive practice is inapplicable
where there are no modal properties that could be inferred to, as is, e.g., the case
in meta-studies.