摘要：
This article traces the philosophical and psychological connections
between causation and the conditional, if...then, across
the two main paradigms used in conditional reasoning, the selection
task and the conditional inference paradigm. It is argued that hypothesis
testing in the selection task reflects the philosophical problems
identified by Quine and Goodman for the material conditional interpretation
of causal laws. Alternative formal theories to the material conditional
only became available with the advent of possible worlds semantics
(Lewis, 1973; Stalnaker, 1968). The relationship proposed by this
semantics between counterfactual and indicative conditionals is
outlined and it is concluded that moving away from the abstractions
of possible worlds proposes a central role for prior knowledge in
conditional inference. This conclusion is consistent with probabilistic
approaches to conditional inference which provide measures of the
strength of a dependency between the antecedent and the consequent
of a conditional similar to those proposed in causal learning. Findings
in conditional inference suggest that people are influenced not
only by the strength of a dependency but also by the existence of
the structural relationship, the broader causal framework in which
a dependency is embedded, and the inhibitory and excitatory processes
like those required to implement Causal Bayes nets or neural networks.
That these findings may have a plausible explanation using the tools
of current theories in causal learning suggests a potentially fruitful
convergence of research in these two areas.
