Right after you wake up from a dream, you seem to be able to remember most if not all of the details. Then, over time, these details fade and often disappear.
Is this a form of regression? Why is it harder to solidify memories of dreams?
Short answer: Because areas of the brain needed for remembering are turned off during dreaming.
The process of converting perception into a memory construct that can be stored is called encoding, and is essentially the same during both wakefulness and sleep: That is, the same factors can hamper or promote successful encoding when awake or dreaming. During sleep, the brain undergoes cyclical changes in which activity in some parts of the brain increases (hyperactive), while other areas decrease in activity (hypoactive). Some areas of the brain associated with memory encoding are hypoactive during dreaming, resulting in most dreams being forgotten.
The most prominent area implicated in memory encoding of episodic memories is the prefrontal cortex (PFC). Activity in this region is correlated with memory in studies conducted in awake subjects, and is similarly correlated with dream amnesia: During dreaming, this area is typically hypoactive (turned off), and its level of activity is predictive of dream recall. Activity in the PFC during sleep may be modulated by levels of acetylcholine, dopamine, and other neurotransmitters.
Nir & Tononi (2010) review the issue:
Theoretically, waking up reactivates the areas of the brain that are turned off during sleep, allowing for memories to be encoded. This process is unreliable, as memory encoding in both awake and sleeping subjects is modulated by attention - also a function of the PFC - so how much and how well dream content is remembered also depends on how much attention is given to it during waking. Without attention, the memory trace fades quickly. This interpretation is consistent with the reduced dream recall in awakenings from deep (slow-wave) sleep, that are characterized by disorientation and lower PFC activity, and with higher dream recall of lucid dreaming, characterized by awareness, control, and higher PFC activity.
Note: The above is certainly an over-simplification of a complex and poorly understood system, but it does suggest that any factor affecting the processes involved in memory encoding and attention (such as certain neurotransmitters, psychoactive drugs, lesions, mood disorders, training, experience, etc) may impact dream recall. On the other hand, NMDA, a neurotransmitter involved in long-term memory consolidation (LTP) - a different activity that is increased (hyperactive) during dreaming - is less likely to be implicated in dream amnesia.
State dependent memory could play a role in quickly forgetting dreams after awakening.
See my question here: What is the scientific term for unexpected, spontaneous dream recall? I ask about a phenomenon where dream recall happens much later potentially weeks or months after awakening.
I would venture to hypothesize that Melatonin might play a role as a state dependent cue that alters the state of memory for recall. Melatonin supplementation can produce extraordinarily vivid dreams that rapidly lose their grandeur upon awakening and subsequent recall.
Personally I take a melatonin supplement at night and can drift off to sleep while recalling dozens of dreams (seems to correlate with melatonin levels rising). My understanding is that Melatonin is suppressed by blue light, so bright lights in the morning may rapidly reduce melatonin concentration and hamper recall.
My suggestion would be that, as with all forms of mental imagery, short-term working memory circuits are responsible for keeping dreamt images active, i.e. in consciousness. Distraction by competing perceptual representations would cause immediate disruption of one's dreamt images; without such distraction, imagery in short-term memory typically starts to rapidly fade after about 8 seconds, unless it's highly relevant to one's immediate situation on waking, and hence promptly acted on.