First, an example. Recall
from an initial generation with a
single live cell the N = 3, S = 2 rule produces the pattern. 



Suppose the rule is replaced by its reflection. How will the
pattern change? In this case, the obvious guess is right. 



So the rule here should produce two lines, one going right, one
going left, right? That is the most frequent quess. Is it correct?
Think a moment, then look at the
answer. 



A few more examples illustrate
the richness of the behavior of onedimensional binary N = 3 CA. 
Changing the rules obviously can have a large influence on the
pattern that evolved. For some automata there is another type of sensitivity:
changing the initial conditions
can have a large effect. 
Not surprisingly, twodimensional
CA also exhibit a rich variety of
patterns. We cannot easily view the spacetime patterns of these. Rather, we present
pictures of a single generation. 
The bestknown of all CA is John Conway's
game of life. 
With the remarkable range of behavior demonstrated by CA, a
natural question is can the behaviors be classified 
If CA behavior can be classified, are there calculations to
predict the behavior? 