Attempts at a physical picture of how living beings develop from unliving m
atter are beset with difficulties due to the low probability of some of the
stages of the process. It is shown that these difficulties arise from the
misconstruction of the term 'coding' and that they are overcome by assuming
that the polynucleotide catalised rather than 'coded' the formation of pro
tein in primary organisms (hypercycles). A realistic scenario including the
emergence of a unified code is considered for such a process. A physical m
echanism of thinking and the development of a necessary evolutionary mechan
ism are discussed, for which purpose the concepts of information, valuable
information, and information generation are analyzed. It is shown that thin
king reduces largely to image recognition. A possible molecular mechanism o
f recognition is considered which is shown to be quite likely to have appea
red in the course of evolution.