%%% File: ./footnotes/Prefrontal_Basal_Ganglia_Working_Memory.tex % \footnote{% % \rawhtml \endrawhtml % The following email exchange builds on an earlier {\urlh{https://web.stanford.edu/class/cs379c/archive/2018/calendar_invited_talks/lectures/04/12/videos/Randall_OReilly_CS379C_04-12-18.mp4}{lecture}} in 2018 by Randy O'Reilly and follow on discussions with Randy and Michael Frank in January and June of 2020. %%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \hrule{} %%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% {\bf{TLD:}} We read several of your papers, listened to your lectures from past classes, reviewed our conversation with Randy back in January and followed the citation trail to track down some of the details that we couldn't find in your work {\endash{}} which isn't to say they weren't there for us to find and we simply missed them. Here are some of the questions that came up in discussions with students that I couldn't answer and couldn't readily find a satisfactory answer for in the literature. A couple of YES/NO answers and one or two references to relevant papers would be most appreciated. {\bf{TLD:}} What's the timing of the "basic cycle" of the basal ganglia including the direct and indirect pathways? What is the average time between BG to PFC/WM transfers? {\bf{RCO:}} 50 msec is the minimum, corresponding to the dominant beta frequency in the BG / PFC circuits ... interestingly ACT-R came up with the same number based purely on psychological fits {\endash{}} see Jilk et al., 2008~\cite{JilketalJETAI-08} for more on that (referred to in attached paper). {\bf{TLD:}} How frequently does the pattern of sensory activity in the posterior cortex get updated in the striatum — with the default answer being "every basic cycle"? {\bf{RCO:}} Direct connections = continuousl, though the STN may act as a bit of a "gatekeeper" such that striatum doesn't "register" the inputs until a relevant top-down PFC context is engaged. {\bf{TLD:}} Does each stripe-like "register" have its own GO/NOGO gating signal, thereby allowing the BG to independently set these signals and initiate transfers to working memory for just the GO-marked registers? Randy seems to have confirmed this in our January meeting but in fact he simply agreed to my interpretation of his comments and my interpretation now seems ambiguous to me. {\bf{RCO:}} This has been the standard PBWM model {\endash{}} see discussion in the [attached] chapter for more on this {\endash{}} not at all a settled question. {\bf{TLD:}} What control do the circuits in the frontal cortex have over the working memory registers that are aligned with those in the striatum {\endash{}} specifically, can the prefrontal and motor cortex circuits prevent the BG from writing to those registers? {\bf{RCO:}} Yes {\endash{}} PFC is major controlling input to its own "stripes" {\bf{TLD:}} And finally, can the BG alter representations in the sensory cortex or are its only targets working memory registers in the frontal cortex? The question originated from a student who read the following paragraph written by Michael in an article on the Dana Foundation website about Parkinson's Disease and was puzzled by the sentence in bold font: % \begin{quotation} % The basal ganglia are a collection of interconnected areas deep below the cerebral cortex. They receive information from the frontal cortex about behavior that is being planned for a particular situation. {\bf{In turn, the basal ganglia affect activity in the frontal cortex through a series of neural projections that ultimately go back up to the same cortical areas from which they received the initial input.}} This circuit enables the basal ganglia to transform and amplify the pattern of neural firing in the frontal cortex that is associated with adaptive, or appropriate, behaviors, while suppressing those that are less adaptive. The neurotransmitter dopamine plays a critical role in the basal ganglia in determining, as a result of experience, which plans are adaptive and which are not. % \end{quotation} {\bf{RCO:}} Yes, BG gates thalamus that projects to frontal cortex, not sensory cortex (though maybe some minor such projections in IT / MTL cortex {\endash{}} not sure about current status of that) {\bf{MJR:}} The sentence in the paragraph your student read was meant to highlight that each FC-BG circuit is a loop {\endash{}} the frontal cortex proposes candidate actions (motor or cognitive) and sends those to the BG which then gates their selection by amplifying the desired one and suppressing the others via disinhibition of that same cortical area that sent the proposal in the first place. This is separate from the fact that sensory (posterior) cortex also projects to both striatum and to frontal cortex, representing the state that then triggers the candidate action generation and contextualizes the striatal valuation. (Though as Randy alluded there is also some anatomical evidence that there could additionally be BG control over higher posterior cortex {\endash{}} my guess is that this can explain visual hallucinations for example when striatal DA is too high.) {\bf{MJR:}} In addition to PFC controlling its own updating in a single loop, our models and data suggest a hierarchical organization such that more anterior PFC-BG loops can provide control over more posterior FC-BG action selection, e.g., when a higher order rule needs to contextualize decision making or during task switching etc. This also facilitates generalization of abstract rules to new sensory contexts. see for example the modeling and imaging work with Badre and Collins.}