With a series capacitor between the (current source?)
and the upper RX terminal, it doesn't look much like a
resistor to me.
The mentioned photoFET optoisolator is only linear for
maybe +/-100mV and then rolls over. Not a good general
purpose "resistor" for anything but true small signal (as
the datasheet itself makes plain).
I wonder about something like CdS photocells, not so
popular now but once nearly the only game in town;
more of a photoconductivity device, possibly with
some decent linearity. Matched pair of these, matched
pair of LEDs (and matched coupling), one in a feedback
loop and one for the resistor role, would be an isolated
variable resistor.
You might also consider "digi-pot" ICs, which would let
you (duh) control them digitally if that's where the
control ultimately comes from, or use an ADC if you
must have an analog control source.