Considerable research indicates that there is a strong tendency for pe
ople observing a social interaction to perceive a given interactant as
increasingly influential or causal as he or she becomes more visually
salient. The present research examined two possible limits on such sa
lience effects in causal attribution. In a situation where one interac
tant is clearly more influential than another, we expected that greate
r salience would increase the estimated causality of the high-influenc
e interactant, but would decrease the estimated causality of the low-i
nfluence interactant. We also anticipated that people who are inclined
to process information in a careful and thorough manner (i.e., those
high in need for cognition) would be unaffected by differential salien
ce. Results revealed the typical salience effect pattern regardless of
whether the observed interactant was highly influential or not, and w
hether the observer was high or low in need for cognition. This resear
ch, then, provides further evidence that salience effects in causal at
tribution are quite robust and generalizable.