pvlv / boa connectivity questions

[rcoreilly]

• OFC / ACC / ALM -> STN ? currently only BLA

• PTp <-> all same areas that CT projects to -- currently trying

[rcoreilly]

• deeper OFC / ACC layers to vsmatrix -- for further contextualizing based on current maint?

[rcoreilly]

From Tom's (@thazy) presentation:

• Price07 defines "orbital" vs. "medial" networks that largely, but not cleanly or exclusively, map onto OFC and ACC, respectively. Orbital is centered on BA13 (more lateral "sensory" OFC; VLO, LO in rats; OFCposUS, OFCnegUS in BOA), while Medial includes BA25 (aversively biased; IL in rat; ACCNegVal in BOA -- there is also a more positively-coding value area ventral to that in BA14?) and BA32 (integrative overall utility; PL in rat; ACCUtil in BOA).

• Should we switch to O vs. M instead of OFC and ACC in the model? or just drop OFC and ACC entirely and go with PosUS, NegUS, PosVal, NegVal, Util as names?

• There are appropriate Motor projections into OFC (13) but not M network -- could give these more of a progress tracking dynamic?

• OFC does NOT receive strong projections from BLA -- how do we reconcile this with Schoenbaum data? which OFC are we talking about? @thazy would be good to track that down more -- this is a cornerstone of the BOA / PLVL architecture.

• OFC represents time independent of effort, while CMA codes for effort: HosokawaKennerleySloanEtAl13 -- time is a more "sensory" value while effort is tied to motor action plans, and requires motor input.

  • We should probably have 2 separate "NegUS" layers: existing NegUS is OFC sensory version and has pool[0] as pure time, and then add ActCost layer that gets direct input from dorsal PFC motor areas (dlPFC & SMA, or ALM in rats). In this case, it may make more sense to retain OFC as sensory US? Just O?

• O vs. M have offset connectivity in MD thalamus. M is more ventral -- perhaps the most ancient, core such area while O is more recent and elaborates sensory US reps that otherwise are not coded in e.g., reptiles etc. Would be interesting to see if there are any implications for VS -> MD projections.

[rcoreilly]

Added the pure time negative US as another neg valence pool in #302 -- this is also good for allowing additional effort to be compared relative to raw time.

[thazy]

Re: the claim by Price07 that amygdala-orbital connectivity is relatively light compared to medial PFC, the short TL;DR answer is that I’ve looked at a bunch of stuff and have satisfied myself that there’s more than enough amygdala-OFC connectivity there to undergird our basic story and all the Holland/Gallagher/Schoenbaum, etc. data is still valid and relevant. Thus, Price07 sorta overstated his case in this regard IMO.

However, the pattern of connectivity to OFC is pretty striking in that the connectivity is densest posteriorly in the agranular insula (Ia subareas that everyone seems to count as "orbital cortex") and then extending both medially and laterally to explicitly orbital cortex in a kind of semi-circle; thus, importantly, the central orbital surface going anteriorly is rather surprisingly devoid of amygdalar connections. This is best illustrated in Figure 9 from the AmaralPricePitkananEtAl92 chapter in the Aggleton-edited ‘The Amygdala’ (pdf now on Zotero) and also Figure 3 from CarmichaelPrice95a.

As this pattern of connectivity would seem to constrain theories of amygdala/OMPFC function -- and the subareas thereof -- with potential implications for BOA, etc., I’m planning to present on all this and some implications, etc. for next week’s show-and-tell — I think there’s a lot of really interesting stuff to discuss. In particular, another chapter from ‘The Amygdala’, Gaffan92 (on Zotero), was a real revelation for me in terms of a theory of amygdala function, and by extension OMPFC function. BTW, @rcoreilly, if you still have the second edition of ‘The Amygdala’ (2000 vintage, I believe), it may have updated chapters by Amaral et al. & Gaffan that might be worth looking at — and/or any other new stuff too. (I only have the 1992 edition.)