Firstly, there is a need to look at the mechanism of dreams. Clearly, the brain has stored memories and images upon which it can draw. It appears to select a theme and construct a story. But without fail, it is disjointed; it falls down on continuity. Characters change, events move from place to place in irrational sequences. However, the common theme is one of a message; the dreams stir our emotions. We all dream, whether we remember the act or not; it is a natural phenomenon and not one borne out of psychosis or abnormality.
The brain is also creative in that it constructs new images from its memories. How often have we seen the bizarre monster; a composite of more familiar animals? But the exact processes that initiate the dream, and the sequence of neural connections that play over the story, remain a mystery. Is the dream pre-programmed? It is assembled, stored and then enacted? Or does it progress haphazardly from a random start? As yet, such questions cannot easily be answered.
Secondly, there is the question of when we dream and the intended effect. During the normal sleeping night, the first phase (the 'alpha' stage) finds the muscles relaxing, blood pressure dropping and the heart slowing. This progresses through two further stages until in our deepest sleep ('delta' sleep) we are immobile and very difficult to wake. This is called 'Non-Rapid Eye Movement' (NREM) sleep. It usually lasts for about 90 minutes and we are not dreaming.
In fact, we dream mostly at the top of our sleep cycles as we near waking, particularly towards morning. This is the so-called 'REM' mode (Rapid Eye Movement) during the 'active' sleep cycles which, as demonstrated by many controlled studies, corresponds to maximum dream frequency. These cycles last only 10 to 20 minutes and the act of dreaming is indicated by flickering eyes as if watching a film behind the eye-lids. There is also increasing oxygen consumption and activation of the adrenal glands. Brain activity suggests great attentiveness; yet paradoxically, the body remains immobile and the muscles are limp. This 'paradoxical' sleep brings near total paralysis, perhaps deliberately to prevent the dreamer physically enacting out the dream content, maybe thrashing out the virtual threat as seen behind the eyes.
We sometimes wake temporarily to feel the emotion; to panic, to wait for calming, or to relish pleasure or excitement. It's then that we can best recall the dream, but it is quite possible that dreams are not intended for memory. In my own studies, I often found that I could remember the act of dreaming, but all hint of the content was lost. Only by using the brief waking interlude to consciously file the dream as a memory could I recall it. Even then, morning would often find that the filing had failed. It's comparable to temporally writing to a computer disc, to be quickly overwritten unless the file is deliberately saved.