Certain events shock us in to action despite the fact that we experience different events of identical or greater magnitude that do not shock us in to action. More technically, the (negative) value, v, of some event that we find acceptable may be greater than some v that we do not. So when we evaluate risk as the probability, P, times v of an event it should be relatively easy to calculate which of two events have the greater risk and subsequently act accordingly.
But as a matter of human psychology, we do not. There are many asymmetries we observe in humans that belie this obvious point. For example, many people think it worth insuring against misfortune whilst they are averse to a quantitatively identical gamble on fortune, that is even where the absolute value of the probability of (mis)fortune times possible loss in the insurance case is equal to the absolute value of the of the probability of gain times the value of the gain in the gambling case. (Insurance companies, by the way, make money because they do not differentiate between the two.)
A fairly concrete example, people are more scared of flying than driving despite the fact that we all know that planes are a safer mode of travel, statistically, than cars. That is, the probability of being injured on a plane times the severity of the possible injury is far lower than the corresponding number for a car.
There are numerous psychological reasons to account for the asymmetries in the types of cases illustrated above. Alan Ryan's paper "Risk and Terrorism" (in Risk: Philosophical perspectives, Tim Lewens, ed; Routledge: 2007.) claims that there are likely evolutionary reasons why we make the judgments we do, but the overriding reason we accept some risk is because of the way it is "framed". Airplane incidents, for example, when they occur, don't just go wrong, they go spectacularly wrong and it is the shock-value that causes us to mis-estimate the probabilities of certain kinds of events. Ryan refers to an event's salience, its size, proximity, and surprise factor that contribute to an event's ability to seize our attention (181).
Terrorists too take advantage of humans' inability to innately miscalculate P(v) by relying on spectacular incidents: attacking the World Trade Center or the Pentagon, a suicide bombing at the wedding of a coreligionist, or a videotaped beheading. Interestingly, most such terrorism accomplishes little. Moreover, groups that employ such tactics make few realistic demands of the governments they attack.
It seems to me that this all speaks to a number of issues including the ability of terrorists to use rational tactics (regardless of the immorality of the tactics or irrationality of their goals), and the fact that we need an approach to counter-terrorism that realizes that these are the tactics and the effects and thus we need to prepare accordingly.