So how does it work?   When one input beam is kept on all the time 
    light exits the hole in the mask when the other beam is off. That is, 
    the hologram which is lit, focus its light to the single spot image 
    which is located right in the middle of the mask hole...thus it exits 
    the transistor through the hole. 

      When the modulated beam is switched on, destructive interference  
    takes place at the hole causing the photons to be dispersed elsewhere 
    on the output square.  Since the entire output square is the mask,  
    and therefore black, it doesn't really matter where the photons land  
    on the mask.  The important thing is that they are blocked anytime 
    both inputs are on...and exit the hole when only one beam is on. 

      Most importantly, the process occurs at the speed of light, and 
    electronic NOT circuits are considerably slower. If fact, one should 
    note that the inputs to an electronic NOT must remain at constant 
    states during the entire switching time. That is, the modulated input 
    must remain on for an entire nanosecond in order for a one-nanosecond 
    switching electronic transistor to produce a useful output.  Not so 
    with the photonic NOT.