Defense specialists in and out of the Pentagon are attempting to focus new attention these days on finding answers to an old problem: how to avoid a surprise military attack.

From the Japanese bombing of Pearl Harbor in 1941, to the North Korean invasion of South Korea in 1950, to the Arab-Israeli wars of 1967 and 1973, to the Argentine grab for the Falkland Islands earlier this year, virtually every important military clash of the past four decades has begun as a military surprise.

And, as a study on "Surprise Attack" published yesterday by the Brookings Institution points out, those attacks almost always were initially successful and almost always were preceded by events and intelligence information that, had more attention been paid, would have given ample warning of what was coming.

In other words, as the study by Richard K. Betts points out, surprise frequently succeeds because political leadership fails to respond properly or promptly to the warning signs.

This may be because the leaders do not believe their intelligence reports, or do not believe the would-be enemy would risk war, or because they fear that their military preparations would escalate a crisis into a war, he says.

This theme was stressed in February by Defense Secretary Caspar W. Weinberger in his annual report to Congress.

"It is sobering to recall how often elaborate warning systems failed to trigger the needed decisions to prepare against surprise attack," he said, adding that "Soviet military doctrine puts great emphasis on deception and surprise."

Because of these factors, Weinberger said "we have to change our policy for reacting to warning. Our forces and those of our allies must be prepared to respond to warning indicators that are highly ambiguous."

Pentagon officials say talks have begun with some NATO allies on ways to improve responsiveness to warning signals as part of a general effort to get governments to pay more attention to the problem. The initial effort, however, was a new NATO study that officials here said did not yield much. Indeed, Betts says in his 300-page study that the possible changes in policy or military posture that would reduce the danger of surprise "are quite limited."

Nevertheless, he argues that NATO, because of its proximity to the potential front lines in Europe, and the United States, which would have to reinforce Europe or send troops elsewhere in the world, both need to think harder about developing new strategies and tactics that will counter surprise by reducing dependence on warning time.

For example, he says NATO might be better off with a forward defense that emphasizes heavily fortified defense positions, using newly constructed fortifications and natural terrain advantages, that could be manned with fewer troops and tanks than current plans provide, and with new high-technology weapons that perform best from fixed positions.

This would reduce the risk of big losses at the outset of an attack, he says, while freeing more troops for use as mobile reserves behind the line for counterattacks. The lack of such reserves has always been a NATO shortcoming.

Betts says such a strategy also would not be provocative because it would not necessarily require movement of troops from the United States on the basis of warning signals alone.

In general, Betts says that as a hedge against surprise, "mobility and flexibility are more important than quantity or mechanical quality because the latter advantages will be moot" if they are wiped out in a surprise strike or cannot be deployed properly.