Let me see if I can help you out
with this because I know how confusing this can be.

The "Assessed Level of Control Risk" and the "Risk of
Assessing Control Risk too Low" are two different
things.

"The Assessed Level of Control Risk" is what
you assess (believe) control risk to be.  If you
assess that control risk is high, you will have
an increased sample size.  That's why there is a
direct relationship.

"The Risk of Asessing Control Risk too Low" is how
much risk you are willing to take that your assessment
about control risk is wrong - Actual control risk is
higher than your assessed (believed) control
risk.  The more risk you are willing to take on
that your assessment of control risk is too low, the
smaller your sample size.

ETA: "Tolerable Misstatement" and "The Risk of
Assessing Control Risk too Low" work the same way.  If
you are willing to tolerate more misstatement,
you have a smaller sample size.  If you are willing to
tolerate more risk that your assessment of
control risk is too low, you have a smaller sample
size.

I really hope this helps!