Oddly, you can have different current paths running at the same time in the same conductor - both AC ones and DC one. It's one of those 'magic' things.
I remember back in school (a long time ago) that the really simply circuit stuff we were doing - battery, cable, lamp, back to battery could share a common piece of cable. You could even swap the polarity of one of the batteries and it still worked.
Lars is quite right about these currents having to go somewhere and they do have to have the return path - but this needn't be a chunk of metal, like a wire. It's a bit tricky to explain - but the concept is important - my work isn't with RF it's with entertainment - and we have grief with getting data to things like lights. We stuff data down a piece of coax, and often the lights will go berserk - spinning around, flashing and doing very odd and totally unwanted things. The problem is reflections. The data (in a current stream) goes down the cable, gets to the very last socket, as the data travels in and out of each light - and has nowhere to go - the paths suddenly stops. (similar to aerials, in a way). Instead of stopping, because the coax we use has an impedance (110Ohms, rather than radio's 50), it acts like a mirror and reflects the current back to the source. In our case, this messes up the data - the lights don't know that they are getting two arrivals of data, and get confused. Our solution is very, very simple. We terminate the cable with a simple 110 Ohm resistor. The mirror effect vanishes, and the lights behave. The cable behaves like a transmission line.
In RF terms a length of coax cable with nothing on the end is called a 'stub'. They have properties - capacitive or inductive, depending on the length of the stub - related to the wavelength. We use stubs all the time in radio - you can short the furthest end, or leave it open depending on what you want to do. They're great for making interference filters, or notching out a very strong local signal if you are a scanner user, for example. In operation, the current IS on the inner and outer conductor, but the inner one is shielded by the outer, hence why the outer one does more damage if it radiates.
The other thing to remember about current is that it's linked to voltage by Ohm's law. As current increases, voltage goes down and vice versa. We're dealing with AC so with any aerial system some parts will be carrying low current with high voltage, and other parts will be low voltage with high current. People do talk about things catching fire, as Alan says. If your aerial has been shortened using coils and other tricks, then one part of it can easily be at a very high voltage - remember the early days spark transmitters that got as far as America? You can get a burn from some aerial designs. Other designs need very thick aerial components because in these ones, the current can be high.
Last thing is to remember that transmission lines don't need to use cable at all. Radar uses waveguide. A square or rectangular section piece of metal. Nothing inside, and the RF bounces it's way along the inside (like optical fibre).
The circuit in the question always exists, or current can't flow - exactly how it returns is the clever bit.
The one thing that has always caused me problems is this notion of how they go both ways at the same time - it seems very wrong, and worse still, the old examples of how electrons work in cables seem to fail. I remember it being explained to me by stuffing marbles into a tube - stuff one in one end and another falls out the other end. Works great with DC, but falls over when we talk RF and reflections!