In  trying  to  refute  the  "Berkeley  answer",  John Searle
imagines a man  carrying out in his  head a program for  conducting a
dialog in  Chinese and notes  that the man  carrying out  the program
might   understand  no   Chinese.   He   concludes  from   this  that
understanding is not a property of programs.

        Here two processes are  using the same hardware.  It  is like
Stevenson's %2Dr.  Jekyll and  Mr. Hyde%1 except  that Dr.  Jekyll is
interpreting Mr. Hyde rather than time-sharing with him.  Were either
phenomenon common, we would not identify a human personality with the
human body  just as  we don't  identify a  computer program  with the
computer itself.  Once we distinguish the two processes  taking place
in the same brain,  the Berkeley answer that the  interpreted process
understands Chinese remains tenable.