Memory, nutrition, motivation, and genes
There are two confusing but potentially important papers in the Jan. 25 2013 Science on long-term memory (LTM) formation in fruit flies:
Pierre-Yves Placais & Thomas Preat. To favor survival under food shortage, the brain disables costly memory. 339:440-441.
Yukinori Hirano et al. Fasting launches CRTC to facilitate long-term memory formation in Drosophila. 339:443-446.
These papers categorize long-term memory formation along three axes.
Aversive vs. appetitive: Actions that the brain interprets as helping it avoid something, vs. actions that help it attain something.
Fasting-dependent (fLTM) vs. spaced training-dependent (spLTM): fLTM is formed in a single learning episode, but only at the time that an organism first obtains food after a long fast. spLTM does not require fasting but requires repeated training.
LTM vs. ARM: Memories that require protein synthesis (LTM) vs. “anesthesia-resistent memory” (ARM), which does not. (The papers don’t explain what ARM might correspond to in humans.)
The relationship between these is unclear, particularly as each of these three axes is claimed at various times to determine whether memory can be learned in a single training cycle (appetitive, fLTM, and/or ARM) or not (aversive, spLTM, and/or LTM). But these things appear to be likely, or at least to be reasonable hypotheses, if these pathways are conserved in humans:
How quickly you learn something depends on how much you’ve eaten recently. You learn most quickly immediately after ending a long fast. Your brain thinks you just learned something that saved it from starvation. (But note that a 1-day fast for a fruit fly could be compared to a human fasting for months.)
How quickly you learn something depends on whether your brain thinks that this knowledge is to avoid something bad (slow learning) or to attain something good (fast learning).
Almost all of the mutations that extend lifespan in organisms from yeast to humans impact the FOXO3a vs. mTORc1 axis (to use the human analogs). Expressing FOXO3a inhibits mTORc1 and extends lifespan in various ways; producing and assembling more mTORc1 inhibits FOXO3a and promotes protein synthesis, growth, reproduction, tissue repair, and immune response. We already know that extending lifespan, in general, is antithetical to building muscle. It may also be antithetical to forming long-term memories. This makes sense.
Learning rate can be increased by expressing or inhibiting proteins involved in these responses. Hirano et al. focus on activating a cAMP-regulated transcriptional coactivator (CRTC) by dephosphorylating it in order to invoke fLTM. They were able to do this and enable flies to learn quickly without fasting followed by feeding.
I’d really appreciate it if somebody would do a literature review and a comparison of the pathways involved to those in humans, and summarize their findings.
So, the picture I’m getting from this is that if I start a cult I should have the followers fast for a suitably long time, and then present them with some profound revelation along with a meal. And if they don’t want to fast, the second-best thing is to have them repeatedly come to the same place and be told profound revelations that will positively influence their lives.
You have successfully predicted the past, yes. Many do this.
Considering that many also predict the past unsuccessfully, I shall take this as a solid, though unexceptional, endorsement.
That’s what I meant it as :-) Sleep deprivation and tricks with food are commonplace. (To the point where the Hare Krishnas being slightly famous for really good food is a bit of a surprise. They’re still into the sleep deprivation, though.)
Training by the U.S. Army and by Tom Brown Jr. both involve a lot of sleep deprivation. The Army doesn’t use it in any systematic way that I can tell; they just think it’s good training because they expect sleep deprivation during combat. I think TBJ does it to cause hallucination and general weak-willedness.
Aversive learning is slower than appetitive learning? That’s possible, but not what I would expect from my reading on anxiety and learned aversions, which can often be traced back to a single traumatic incident. Maybe this is just selection bias (EDIT: It was.), or the relative amount of badness is incomparable, but that seems like those results ought to get reconciled somehow.
The animal training book Don’t Shoot the Dog states that reinforcement-oriented clicker training is substantially faster and more persistent than aversion-based alternatives. Even for cases where you’re training an animal not to do something, the author recommends finding a way to make use of reinforcement-based training somehow… e.g. train a behavior incompatible with the one you want to discourage. “[Punishment is] everybody’s favorite [method for getting rid of undesired behaviors], in spite of the fact that it almost never really works.”
Oh, of course. Positive reinforcement in general is stronger, which I would have noticed if I hadn’t been primed by reading about anxiety recently, which suggests I should institute some sort of debiasing exercise when I recognize the risk of selection bias rather than just announcing it.
Wouldn’t you want to learn the thing right before breaking your fast?
The experimental time resolution isn’t that high. They’re comparing things learned “when” breaking the fast vs. things learned 4 or 6 hours earlier or later.
It’s interesting that the idea of learning influenced by emotional state goes way back. There is Talmudic reinterpretation of “But my wisdom remained with me” to “wisdom that I learned in anger stood with me”. I’m not sure if there is epistemic truth to this, but it’s an interesting thought, and I wouldn’t be surprised if it’s not totally wrong; the Talmudists studied pretty damn hard.
This is a well-established effect.