Great Qs. I appreciate the conversation. There’s a lot to unpack. I’ll start with that word: computation.
Upon reflection, I use the term computation in two ways. The first is your way - I don’t believe that orientation tuning exists in V1 because the brain optimized explicitly for orientation, but in service of one or more bigger goals (object recognition, etc). But I still say that ‘V1 computes orientation’.
But! once we discovered that something slightly more mysterious we were interested in (memorability: some images are more memorable than others) was an epiphenomenon that happens in service of a bigger goal (https://pubmed.ncbi.nlm.nih.gov/31464687/), I stopped using ‘computes’ to refer to it. This thing seems to emerge as a consequence of a system wired up to see (it didn’t have to be that way). I’m still interested in how the brain ‘gives rise to it’ and to use that behavior as leverage to probe something bigger (memory), but I’m no longer seeking how the brain ‘computes’ memorability; saying that seems misleading to me. In other words, there are ways in which ‘computes’ implies intention that the thing of interest is at least a benchmark or approximation of something useful (not a byproduct).
I suspect I’m not the only one who flips between these two usages (and it’s helpful for me to know I do).
Back to mood, as far as I understand it, most emotion researchers don’t think that the brain is optimized to compute ‘mood’ but that it exists in service of a larger goal (like approach/avoid > pain/pleasure ….). So a research program focused on the computation of mood seems off from the outset in that same sense that memorability is.
Of course understanding where and how the brain reflects mood (dynamically) will be part of any research agenda on this topic. How one interprets that data in terms of what the brain is computing (and whether computation is even the right way to think about it) is the open question.
One way pursuing mood could be misleading is argued by this camp:
https://www.nature.com/articles/s42003-022-04324-6
They argue that emotions as we experience them are misleading depictions of what happens in the brain (and so it doesn’t make sense to chase ‘mood’ or ‘fear’ in isolation).
Moreover, the metaphor they advocate for is a recipe (not a machine).
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2763392/
Jury is still out about the best way to think about it. But I do agree with Barrett that our metaphors drive our imaginations and thus our research programs (and thus we should be thoughtful about them). Ergo, I worry about jumping in to explore anything that the brain does as ‘computation’ for fear of getting stuck in a misleading local minimum; I think there’s good reason to think carefully about 1) how we define what the brain does and 2) the type of thing the brain is to do it.
