neuroscience - Long-term-potentiation and memory. Where do we stand?

Really good questions. As the guy who brought up LTP/LTD in the question you referenced, I thought I would weigh in.

There is the traditional definition of LTP/LTD as an increased/decreased synaptic efficacy at a single synapse or in a single cell. As you've noted, this is unlikely to be the only phenomena underlying memory and sometimes it's hard to see how some of these mechanisms can result in memory on behavioral timescales.

Let me propose, therefore, that the term long-term plasticity is more relevant these days, as it can refer to a variety of mechanisms that relate to the ability of the nervous system to change in stable ways over time. Physiological mechanisms involving changes in protein expression include traditional LTP/LTD at single synapses, but also homeostatic plasticity and long-term changes in intrinsic excitability where the tendency of the cell to fire changes independent of changes in synaptic weighting. Some structural mechanisms include the growth of new synapses, new spines, and new neurons--synaptogenesis, spinogenesis, and neurogenesis.

In the end, it is all of these mechanisms (and probably more) at play. Note, for instance, that the plasticity may move through structural changes in the system. This means that the lifetime of LTP in one cell or at one synapse does not necessarily have to be the same as the lifetime of the memory itself. All that said, I think all plasticity mechanisms ultimately reduce down to a change in the ability of an input to elicit an action potential somewhere in the brain (known as EPSP-spike coupling). This is likely to be the basic underlying mechanism of memory.