I wrote four paragraphs about different interpretations of touch spells before actually reading the SRD. Silly me. Those paragraphs are gone forever now.
If the character touches anything with the character's hand while holding a charge, the spell discharges. If the character casts another spell, the touch spell dissipates.
This suggests that the charge is automatically discharged when the caster touches something, that the caster's will doesn't enter into the equation. Obviously, air particles must not be sufficient to discharge a touch spell, and players or DMs who suggest otherwise need to write 500 times on the blackboard, "Magic is not science."
If air particles aren't sufficient to discharge a touch spell, is a figment sufficient? I'd argue that it's not: according to the rules, a figment has no material reality.
Because figments and glamers (see below) are unreal, they cannot produce real effects the way that other types of illusions can.
I think in this case the sage's ruling will produce inconsistencies. What if, for example, a blind caster walks through a figment while holding a touch spell -- does it discharge? Can a caster with major image create a swarm of flies that make it impossible for anyone within the illusory swarm to hold a touch attack charge? What if someone within the swarm has true seeing cast?
Conversely, if I am convinced that my opponent is an illusion, can I punch him while holding a touch spell without discharging it?
It is more consistent, I think, to rule that figments and glamers cannot cause touch spells to discharge: this keeps the discharging out of the caster's control, as it should be.
Daniel