|
@rob_gulli | |||||
|
#tweeprint! Have you ever wanted to learn about spatial and non-spatial hippocampal function, especially in primates? We look at single neurons + ensemble activity across tasks during virtual navigation in a new paper with @JulioMTneuro in @NatureNeuro nature.com/articles/s4159… 1/
|
||||||
|
||||||
|
Rob Gulli
@rob_gulli
|
23. pro |
|
I started my thesis wanting to study memory in the hippocampus, inspired in part by Brenda Milner & Wilder Penfield. I quickly learned the complexity of this story, with conversation over the primacy of spatial versus non-spatial mnemonic function in the hippocampus 2/
|
||
|
|
||
|
Rob Gulli
@rob_gulli
|
23. pro |
|
It's difficult to study spatial and nonspatial activity in single neurons (but see great expts by @SuthanaLab, @ArneEkstrom1, @neurojosh, @QasimEtal, @KahanaMichael). For a new approach, we trained monkeys to complete two behavioral tasks in a single VR envt. 3/
|
||
|
|
||
|
Rob Gulli
@rob_gulli
|
23. pro |
|
One task required just cue-guided navigation, inspired by classic studies of rodents foraging for sugar pellets in an open arena youtu.be/aWvheMzxMJo 4/
|
||
|
|
||
|
Rob Gulli
@rob_gulli
|
23. pro |
|
Subjects also completed an Associative memory task in the same environment, which required learning a new set of context-dependent object values every day youtu.be/RHx9Lw65oDw 5/
|
||
|
|
||
|
Rob Gulli
@rob_gulli
|
23. pro |
|
Figs 1 & 2 of the paper show single-neuron changes in spatial selectivity across tasks. In Fig. 3, we ask whether coding of spatial position generalized across tasks. 6/ pic.twitter.com/X2MRecqo80
|
||
|
|
||
|
Rob Gulli
@rob_gulli
|
23. pro |
|
The ensemble could support decoding of position in the maze in each task, but the code for space did not generalize across tasks. There are important nuances and implications here, see the paper for full detail. 7/ pic.twitter.com/o9YmpyAoT9
|
||
|
|
||
|
Rob Gulli
@rob_gulli
|
23. pro |
|
The lack of stable coding across tasks was striking in the portions of the maze that were especially relevant in the Associative memory task. So, we next examined non-spatial, task-specific activity during the Associative memory task. We looked for perceptual & mnemonic coding 8/
|
||
|
|
||
|
Rob Gulli
@rob_gulli
|
23. pro |
|
We found that hippocampal neurons encoded objects, and interaction with context as both perceptual and mnemonic representations. Neural activity could be used to read out specific trial type in the Associative memory task as well. 9/ pic.twitter.com/9XqVqP7Ulx
|
||
|
|
||
|
Rob Gulli
@rob_gulli
|
23. pro |
|
We showed that hippocampal neurons could be used to decode space in primates during virtual navigation. However, representation of space is not stable across tasks. Task-specific activity that looks spatial can be tuned to nonspatial parameters that are relevant for behavior. 10/
|
||
|
|
||
|
Rob Gulli
@rob_gulli
|
23. pro |
|
So what does this mean for our understanding HPC function?
ATM, I believe that neuropsychological theories (memory, space) contain less explanatory power than computational theories (cc @StefanoFusi2 @neuro_kim @jcrwhittington @behrenstimb @ProfData @Rob_Mok @tyrell_turing) 11/
|
||
|
|
||