Christine,
You can try Bob's approach, but if you want to stick with 3dRegAna, your regression model is
Y = b_0 + b_1 * X + b_2 X^2 + e
> We would like to be able to say that activity in brain area A fits a linear
> function, better than it fits a quadratic function.
You would be able to see in such a region b_1 is significantly different from 0 while b_2 is not.
> Conversely, we would like to be able to say that brain activity in area B fits a
> quadratic function, better than it fits a linear function.
In this case you should simply have b_2 significantly different from 0.
HTH,
Gang