It threw me for a moment how this worked without having the action apply to one visible thing- surely the action would fail the basic accessibility rule when trying to recall something which couldn’t be touched? Then I recalled @Zed having pointed out some time past that the basic accessibility rule does not apply to rooms or regions.
Non-room objects of course don’t make it as far as an action, being spat out by the parser with ‘That noun did not make sense in this context’