If you are using the opamps output pin as a voltage output (not as the Howland) , then here is a way of analizing it.
First, assume R3 is infinite. Then it degenerates into a conventional non-inverting amplifier, where for a finite R4, the gain is 1+R2/R1.
Now start making R3 smaller. Because R3 provides positive feedback, the overall gain of stage increases. There is a limit to how small R3 can become without making the positive feedback greater than the negative feedback via R1/R2.
The place where things go wierd is when R3/R4 = R2/R1.
I'll let you do the equations...
btw, there is a "homework" forum for questions like this...