No need for heating or cooling. As already mentioned previously, it is a effect of modulating the source impedance of receivers input amplifier detected as slight amplitude modulation. The 'original' part here is that modulated source is 'connected' to receiver using closely coupled horn antenna link instead of transmission line.
I would not agree with this this statement:
The thermal noise in their receiver would be the same if not more otherwise, then add path loss....
They use relatively high gain antennas, which effectively attenuates signals, including ambient thermal noise from directions other than bore-sight. See classic integral formula here:
https://www.antenna-theory.com/basics/temperature.phpBy placing such antennas directly facing each other, most of the above integration in case of RX antenna would be dominated by aperture of TX antenna, and that means that we can actually cool and heat the TX resistor, and that would also be detected as amplitude change even if ambient thermal noise is say 300 K. As matter of fact they did such experiment (see their paper) using liquid nitrogen cooled resistor.
All this is nothing new, papers/'discoveries' like that slip through all the time.