Analysis of memory in learning is often framed in terms of probabilities, e.g. the chance you'll recall something on a test, which might decline over time.
This does describe the data, but it's funny how alien this is to how memory feels. It does not *feel* like a binomial draw!
Conversation
I often feel like I can "almost" remember something (but then can't), or like I "easily" remember something. Experimentally, these judgments do map to recall probabilities, but is it actually a stochastic process?
8
1
27
I tend to identify more with models of forgetting which aren't truly stochastic, but which can be *described by* probabilities at a macro scale.
For instance, encoding variability theory suggests that recall depends on overlap between mental state at encoding and recall.
4
2
16
Because mental states will naturally vary as you move about and think different thoughts, this naturally gives rise to a memory behavior you can describe with probabilities, but it's not like your brain is rolling dice. The "randomness" comes from behavior & the environment.
1
12
(Encoding variability theory is also a nice potential mechanistic explanation for the spacing effect: when cramming you re-encode an item several times with the ~same mental state; when spacing, you encode it against different mental states, thus more chance for future overlap)
Replying to
I guess this is not unique to models of memory. For instance, language models often work by estimating the probability that the next word is "banana" or whatever. It's not like prose is stochastic, but probability is a useful way to capture co-occurrence.
1
7