Events. Semantically: time-ordered list of future values. Instances:

• Monoid: mempty is the event that never occurs, and e mappend e' is the event that combines occurrences from e and e'.
• Functor: fmap f e is the event that occurs whenever e occurs, and whose occurrence values come from applying f to the values from e.
• Applicative: pure a is an event with a single occurrence at time -Infinity. ef <*> ex is an event whose occurrences are made from the product of the occurrences of ef and ex. For every occurrence f at time tf of ef and occurrence x at time tx of ex, ef <*> ex has an occurrence f x at time tf max tx. N.B.: I don't expect this instance to be very useful. If ef has nf instances and ex has nx instances, then ef <*> ex has nf*nx instances. However, there are only nf+nx possibilities for tf max tx, so many of the occurrences are simultaneous. If you think you want to use this instance, consider using Reactive instead.
• Monad: return a is the same as pure a (as usual). In e >>= f, each occurrence of e leads, through f, to a new event. Similarly for join ee, which is somehow simpler for me to think about. The occurrences of e >>= f (or join ee) correspond to the union of the occurrences (temporal interleaving) of all such events. For example, suppose we're playing Asteroids and tracking collisions. Each collision can break an asteroid into more of them, each of which has to be tracked for more collisions. Another example: A chat room has an enter event, whose occurrences contain new events like speak. An especially useful monad-based function is joinMaybes, which filters a Maybe-valued event.

Reactive behaviors. They can be understood in terms of a simple model (denotational semantics) as functions of time, namely at :: BehaviorG t a -> (t -> a).

The semantics of BehaviorG instances are given by corresponding instances for the semantic model (functions). See http://conal.net/blog/posts/simplifying-semantics-with-type-class-morphisms/.

• Functor: at (fmap f r) == fmap f (at r), i.e., fmap f r at t == f (r at t).
• Applicative: at (pure a) == pure a, and at (s <*> r) == at s <*> at t. That is, pure a at t == a, and (s <*> r) at t == (s at t) (r at t).
• Monad: at (return a) == return a, and at (join rr) == join (at . at rr). That is, return a at t == a, and join rr at t == (rr at t) at t. As always, (r >>= f) == join (fmap f r). at (r >>= f) == at r >>= at . f.
• Monoid: a typical lifted monoid. If o is a monoid, then Reactive o is a monoid, with mempty == pure mempty, and mappend == liftA2 mappend. That is, mempty at t == mempty, and (r mappend s) at t == (r at t) mappend (s at t).