Alternative Possibilities and Frankfurt Cases 107can’t plausibly have this explanatory role. One might, for instance, imagine that
Jones will decide to kill Smith only if Jones blushes beforehand, and that Black’s
device will activate only if Jones does not blush by a certain time (Fischer, 1982,
1994). Then Jones’s failure to blush by a certain time might be the alternative
possibility that would trigger the intervention that causes him to kill Smith. Sup-
posing that Jones acts without intervention, we might well have the intuition that
he is blameworthy despite the fact that he could not have done or chosen other-
wise, or formed an alternative intention. He could have failed to blush, but as
Fischer argues, such a flicker is of no use to the leeway theorist since it is too
“flimsy and exiguous” and “insufficiently robust” to play a part in grounding
moral responsibility.
But what is it for an alternative possibility to be robust, and why would the
robustness of an alternative possibility be crucial? As we’ve argued, the intuition
underlying the alternative- possibilities requirement is that if, for example, an
agent is to be blameworthy for an action, it is crucial that she could have done
something to avoid being blameworthy. If having an alternative possibility does
in fact play a role in explaining an agent’s moral responsibility for an action, it
would have to be robust in at least the sense that as a result of securing that
alternative possibility instead, the agent would thereby have engaged in a volun-
tary undertaking whereby she would have avoided the responsibility she actually
has for the action (McKenna, 1997; Mele, 1996; Otsuka, 1998; Pereboom, 2001;
Wyma, 1997). If Jones had failed to blush, he would not in this way have volun-
tarily avoided responsibility for killing Smith. So failing to blush in the above
scenario does not count as a robust alternative.
Robustness also has an epistemic component. Suppose that the only way Joe
could have avoided deciding to take an illegal deduction on his tax form—a
choice he does in fact make—is by voluntarily taking a sip from his coffee cup,
for unbeknownst to him, the coffee was poisoned, so a sip would have killed
him immediately; or, for another version, it is laced with the drug that induces
compliance with the tax code (Pereboom, 2000, 2001, 2014; cf. Moya 2006: 64,
who constructs an example of this form). In this situation, Joe could
have behaved voluntarily so as to preclude his choice to evade taxes, which
would have precluded the blameworthiness he actually incurred. But whether
he could have voluntarily taken the sip from the coffee cup is intuitively
irrelevant to explaining why or whether he is blameworthy for his decision.
What’s missing is that Joe has no inkling that taking the sip would render him
blameless. This motivates the epistemic component of the robustness condition:
If Joe were blameworthy because he has an alternative possibility in this situ-
ation, it must be that he in some sense understood that or how it was available
to him.
On the assumption that in any situation taking the morally best option fully
available to the agent renders her blameless, here is a proposal (a substantial
necessary condition—not a sufficient condition) that accommodates all of these
concerns. We state the criterion in terms of blameworthiness by contrast with
moral responsibility more generally: