The way I understand it, memory goes through calcium influx induced LTP (Long Term potentiation) for short term storage. Long term storage requires protein synthesis. (#1).
However, a meta-analysis (#2) shows that:
a procedural memory may never become fully stable and instead remains vulnerable to interference. When exposed to task B, participants may have retrieved the procedural memory associated with task A and modified it, ‘overwriting’ the memory for task A with information relevant to task B............ It implies that learning a new skill leads to the automatic destruction of another skill. It would never be possible to have skill in more than one task! Our ability to acquire multiple skills may depend upon having contextual cues available to signal the switch from one task to another.
This indicates that procedural memory never goes through protein synthesis. The problem is, procedural memory can be divided into three parts: 1. Motor, 2. Perceptual, 3. Cognitive (#3). We all know that motor skills are not volatile after being practiced. Only something like Parkinson's disease can damage motor skills (#4) while perceptual skills seem even more secure. Cognitive procedural memory alone seems very vulnerable, but I don't know why this is an exception.
Is there any theory that explains how our brain can store motor skills without any apparent LTP and protein synthesis phases?
#1. Biopsychology by John P. J. Pinel 6th ed. P. 285
#2. http://www.cell.com/current-biology/pdf/S0960-9822(04)00937-6.pdf
#3. http://medicalfinder.jp/doi/abs/10.11477/mf.1416100315
#4. http://www.tandfonline.com/doi/abs/10.1080/01688639008400978