In metacontrast masking, the effect of a visual mask stimulus on the percep
tual strength of a target stimulus varies with the stimulus-onset asynchron
y (SOA) between them. As SOA increases, the target percept first becomes we
aker, bottoms out at an intermediate SOA, and then increases for still larg
er SOAs. As a result, a plot of target percept strength against SOA produce
s a U-shaped masking curve. Theories have proposed special mechanisms to ac
count for this curve, but new mathematical analyses indicate that it is a r
obust characteristic of a large class of neurally plausible systems. The au
thor describes 3 quantitative methods of accounting for the U-shaped maskin
g effect and analyzes 4 previously published mathematical models of masking
. The models produce the masking curve through mask blocking, whereby a str
ong internal representation of the target blocks the mask's effects.