It really depends on the complexity of the system. All of your detectors and activating switches should be wired so that they are normally closed when the system is normal. That way to activate the alarm, the circuit has to be broken. This is done for several reasons. First, if something damages one of the loops, the system can sense there is a problem, rather than discovering the problem when the system fails to work. Secondly, if the fire were to start in an area away from the sensor and burn through the wires, it would activate the system, or at least indicate that there is a problem. Often, when a sensor or alarm pull is tripped, it switches from its low impedance state to a higher impedance state, activating the alarm. An open state is used to indicate a problem with the system, but not necessarily an activated alarm (though often it is treated as such for safety reasons). If you are doing this circuit for a demonstration, that's perfectly fine. If this is meant to be used as a real system, I highly recommend that you seek a professional service regarding this, as what you may end up building will probably not meet fire code. This could end up putting you in serious legal liability trouble should it malfunction.