Nicole Rust on Nostr: Trying to wrap my head around what makes for a good model (of something complicated): ...
Trying to wrap my head around what makes for a good model (of something complicated):
Following on my enthusiasm for a certain type of explanation of "mood" (summarized in this post: https://neuromatch.social/@NicoleCRust/111262116419902198) I'm trying to wrap my head around my lack of enthusiasm for a different type of explanation of the same thing. I'm not here to dump on anyone's work and I suspect that it's probably largely a matter of taste (and I anticipate the other model will resonate better with some other folks). Really just trying to wrap my head around differences in "explanations" of complicated things in ways that invite conversation.
In explanation 1 (which resonates with me): One big goal of the brain is to find good things (and avoid bad ones) and it does this, in part, via reward-based learning. Here, mood is a running average of unexpected rewards and it is formalized in models of reward based learning to demonstrate how "mood" makes learning more efficient. The gist: mood informs us about how things (like our environment) are changing and motivates us to act accordingly. One thing I appreciate about this approach is that it facilitates the mathematical formalization of slippery things (like mood) into models that can be tested, and those models inform not only what mood is but also what it's good for.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4703769/
In explanation 2 (which I struggle a bit more with): One big goal of the brain is to maintain stability in the face of a changing environment. To do that, it has to make predictions and act to minimize its uncertainty (aka minimize free energy via active inference). In this framework, acute emotions signal uncertainty about the environment and mood is a hyperprior that furnishes a higher-level prediction about the value of lower-level emotional states. The purpose of mood is to convey confidence about the consequences of actions. One thing I struggle with is that after I read those words I don't feel that I have a much better sense of what mood is or what it's for than I did before I read them; it's not that I don't understand this framework at large; it's more that I do not experience an "aha" when I read it.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6340107/
What I suspect one difference is: I anticipate that mood is not just about reward-based learning and explanation 1 is really just a slice of what mood is about; in contrast, explanation 2 is more expansive (it's something akin to a theory of everything). But I wonder if this coverage is the same thing that makes explanation 2 a bit too vague for me to appreciate.
Thoughts?
Following on my enthusiasm for a certain type of explanation of "mood" (summarized in this post: https://neuromatch.social/@NicoleCRust/111262116419902198) I'm trying to wrap my head around my lack of enthusiasm for a different type of explanation of the same thing. I'm not here to dump on anyone's work and I suspect that it's probably largely a matter of taste (and I anticipate the other model will resonate better with some other folks). Really just trying to wrap my head around differences in "explanations" of complicated things in ways that invite conversation.
In explanation 1 (which resonates with me): One big goal of the brain is to find good things (and avoid bad ones) and it does this, in part, via reward-based learning. Here, mood is a running average of unexpected rewards and it is formalized in models of reward based learning to demonstrate how "mood" makes learning more efficient. The gist: mood informs us about how things (like our environment) are changing and motivates us to act accordingly. One thing I appreciate about this approach is that it facilitates the mathematical formalization of slippery things (like mood) into models that can be tested, and those models inform not only what mood is but also what it's good for.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4703769/
In explanation 2 (which I struggle a bit more with): One big goal of the brain is to maintain stability in the face of a changing environment. To do that, it has to make predictions and act to minimize its uncertainty (aka minimize free energy via active inference). In this framework, acute emotions signal uncertainty about the environment and mood is a hyperprior that furnishes a higher-level prediction about the value of lower-level emotional states. The purpose of mood is to convey confidence about the consequences of actions. One thing I struggle with is that after I read those words I don't feel that I have a much better sense of what mood is or what it's for than I did before I read them; it's not that I don't understand this framework at large; it's more that I do not experience an "aha" when I read it.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6340107/
What I suspect one difference is: I anticipate that mood is not just about reward-based learning and explanation 1 is really just a slice of what mood is about; in contrast, explanation 2 is more expansive (it's something akin to a theory of everything). But I wonder if this coverage is the same thing that makes explanation 2 a bit too vague for me to appreciate.
Thoughts?