In logic, reasoning from known quantities proceeds in two directions: Deductive reason goes inward from the known to what can be assumed because it is in the same subset: "It is a fact that all birds have feathers; a penguin is a bird; therefore penguins have feathers." This logic goes from the general to the specific.  If we ever found a bird that did not have feathers, we would have to change our statement of fact.

Inductive reasoning goes outward from what is known to what can be speculated: the ten nuts I have collected from this tree were all acorns; therefore, the next nut I collect from this tree will (probably) be an acorn. This logic goes from the specific to the general. While from these examples it would appear that inductive reasoning is less reliable, it is the kind of reasoning that moves science forward.  By observing many examples in a set, the scientist then forms an inductive conclusion about the whole set. "All the stars we have observed give off visible light; therefore we conclude that all stars give off light."  You can see that if a star is ever found that does not give off visible light, we must change our premise--our assumption will no longer be "logical."