At the end of his review on the cell biology of synapse formation (2021), Südhoff claims that

there is scant evidence that long-term plasticity per se is physiologically important for a behaviour.

This lack of evidence - in the form of specific "manipulations of some molecules with multifaceted roles that happen to affect LTP" -, he continues,

has not curtailed speculation that LTP is involved in memory, drug addiction, and scores of other human brain activities.

I always considered it a dogma that LTP lies at the heart of memory etc. Does Südhoff's view have many other proponents? And how could the standard view ("long-term memory = long-term potentiation") be so strong without evidence? What is it based on, then? Finally: What are alternatives to LTP to underlie memory?

  • $\begingroup$ Though he has had a respected career including a Nobel prize, more recently concerns have been raised about his body of work: science.org/content/article/… Personally, I find that paragraph about LTP to be pretty ridiculous, as many papers show influences of LTP on behavior and mechanisms of LTP are some of the most studied in neuroscience. $\endgroup$
    – Bryan Krause
    Commented May 29 at 13:19
  • $\begingroup$ But how could he be ridicously wrong? He knows more about synapses (and thus LTP?) than almost everyone on earth. And the article was peer-reviewed in a well-reputed journal. $\endgroup$ Commented May 29 at 13:55
  • 1
    $\begingroup$ Peer review doesn't mean flawless, it just means some peers were not sufficiently critical of the paper for an editor to reject it. I could easily see peer reviewers not focusing on some throwaway language that isn't central to the paper itself. Südhoff studies synapses, that doesn't mean he studies LTP. I would summarize his argument as "the stuff I study is more important than the stuff they study" - I'd wager that pretty much every scientist thinks this to one extent or another, just like everyone thinks their own children are the best. $\endgroup$
    – Bryan Krause
    Commented May 29 at 13:59

1 Answer 1


I will not comment on the particular author's views and answer the question in the title.

There are five conceivable theories for neural memory (following Hebb (1)):

  1. Memory is sustained by reverberations of neural activity which persist after the stimulus disappears.
  2. Memory takes place through the creation and elimination of synaptic structures, which re-wire the neural circuits.
  3. Memory subsists by adjusting the size of existing synaptic structures, which re-wire the neural circuits.
  4. Memory is internalized by cells through homeostatic adjustments in their molecular machinery.
  5. Memory is simply a re-arrangement of the connections between cells through microscopic movements of the synaptic elements (such as axonal boutons and dendritic spines).

An early proponent of 1 was Lashley. Today, 1 is widely thought largely incompatible with existing evidence, although it can arguably play a role in the formation stage of memory. Theory 2 consists of the formation and elimination of new synaptic contacts. In most cases studied, i.e. in the cortex and the hippocampus, it appears that this comes down to the formation and elimination of dendritic spines. Long-term potentiation (LTP) and long-term depression (LTD) come into play with theory 3. Homeostatic regulation can refer to many different processes, such as changes in synaptic or non-synaptic distribution of some resource such as mRNA or actin, or changes in the distribution and electrical properties of particular types of ion channels and receptors. Theory 5 is largely incompatible with the apparent mobility allowed in real brains as shown by microscopic evidence.

Theories 2, 3, 4, and 5 are not incompatible with each other. Hebb suggested all four possibilities and only discarded theory 1. Much earlier, Tanji and Cajal were fighting a similar fight, Tanzi claiming that connections between cells change because the cables change in size (synapses were not discovered yet), while Cajal claiming that new structures must form --later, he changed his mind on this issue.

The association between LTP and memory is overwhelmingly the most popular theory currently in neuroscience and no one could reasonably discard its existence. However, there is evidence that spine formation and elimination can also play a more limited role in the re-wiring of neural circuits. It is still an open question to what extent and with what temporal features such re-structuring is coordinated through active processes that could resemble LTP (2), although it is safe to say that activity --thus, likely LTP/LTD-- is involved at some stage (3). On top of all that, homeostatic plasticity (theory 4) is generally thought to constantly take place, driven by less specific conditions (4).

1: Hebb, D. O. (2002). The Organization of Behavior. Routledge. https://doi.org/10.4324/9781410612403
2: Under what conditions do dendritic spines form?
3: Hayashi-Takagi, A., Yagishita, S., Nakamura, M., et al. (2015). Labelling and optical erasure of synaptic memory traces in the motor cortex. Nature, 525(7569), Article 7569. https://doi.org/10.1038/nature15257
4: Do neurons that *don't* fire together unwire?

  • 2
    $\begingroup$ Nice answer. To connect it back to the paper OP is reading, when Südhoff says "long-term plasticity per se" he means (3); the paper overall is focused on (2). I agree that (2) through (4) are all broadly thought of as critical for understanding long-term memory over time; (1) remains the best model for working/short-term memory though certainly has influences on the others because they are activity-dependent. $\endgroup$
    – Bryan Krause
    Commented May 30 at 14:14

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .