Piccione and Rubinstein's (1995) absentminded driver example seems to
have touched a nerve. It is hard to believe that a game with one playe
r and one information set could inspire such an outpouring of ideas fr
om so many eminent and intelligent people! In this note, I attempt to
reconcile a few of these arguments. With some embarassment about it, I
will restrict myself to three rather transparent observations. Briefl
y, the first observation is that Gilboa (1996) and Aumann, Hart, and F
erry (1996) (henceforth AHP) are making essentially the same argument.
The second is that Gilboa's form of this argument provides what strik
es me as a very clear ''resolution'' of the paradox noted by Grove and
Halpern (1995). The third is that much of the discussion has taken as
given a view which I believe confuses some issues-namely, that there
is one true way to model the perceptions of an agent with imperfect re
call.