John Greco is a Reliabilist. Like other Reliabilists, he claims* that one cannot have knowledge through “strange and fleeting processes.” As an example of such a phenomenon, he points to an illustration in which one’s eyesight works only intermittently. Since it works only intermittently it is “strange and fleeting” and therefore unreliable.

Stewart Cohen** disagrees with Greco’s conclusions based on the case of intermittent eyesight and uses two scenarios to illustrate his point of disagreement. Assuming that eyesight is part of our cognitive equipment which is responsible for the formation of our visual beliefs, consider my slightly modified versions of those illustrations.

The first scenario is one in which an agent’s eyesight works (i.e. it is “turned on”) but produces an equal number of false beliefs as it does true beliefs. The second scenario is one in which 90% of the time, the agent’s eyesight doesn’t work (i.e.it is “turned off”). For example, 90% of the time, the agent forms no visual beliefs at all. However, when the agent’s eyesight is “turned on”, it almost always leads to the production of true visual beliefs. Cohen believes that we should credit the agent in the second scenario with knowledge but not the first. However, he seems to chalk this judgment up to intuition. He says, “By my lights, these visual beliefs could still be instances of knowledge despite the fact that the process is not not stable in the relevant sense, i.e., it operates in an irregular (intermittent) fashion.” While I agree with Cohen, I think more can (and should) be said.

At this point a principled account of things seems appropriate. I tentatively propose the following principle.

[P]: Knowledge-relevant luck can supervene on cognitive processes that have the potential to produce beliefs; it cannot supervene on cognitive processes that lack this potential.

If [P] is right, we don’t, for obvious reasons, have to worry about knowledge-relevatn luck in the second scenario. Or do we? In order to consider this, let’s go back to our scenarios.

Now I agree that the first agent does not have visual knowledge. This is motivated by the fact that the agent’s true visual beliefs are largely the product of luck; she could have easily had a false belief. However, in the second scenario, things aren’t so simple. As I have already stated, when the agent’s eyesight is “turned on” it regularly produces true beliefs; it is a reliable process/mechanism. However, this fact doesn’t entail that it isn’t a matter of luck when exactly the process/mechanism gets turned on or off. So, in this this case, it is still possible that the agent’s true visual beliefs involve some type of luck.

But it isn’t obvious to me that the luck involved in the second scenario is such that it undermines knowledge. At this point one might begin to distinguish various types of luck and consider how each intervenes on the knowledge-relevant features of agents and their environments. Duncan Pritchard and John Greco have made steps in this direction. If you’re interested, I recommend you Google their names and start reading.

Best,
MLN

* “Agent Reliabilism” Philosophy and Phenomenological Research. Vol. LXVI, No. 2, March 2003
** “Greco’s Agent Reliabilism” Philosophy and Phenomenological Research. Vol. LXVI, No. 2, March 2003