TwistingTwists.github.io

# The GPS of Brain by John ‘O Keefe

Understanding how Brain helps in navigation, anatomical perspective.

• Famous cab drivers of london have BIGGER hippocampus. (where place cells reside.)
• When people stop being cab drivers, the size of hippocampus reduces!

[personal comment (PerCom)]– things exist only to the point they are useful (or they feel love? :p) => Keep doing it daily. Albeit little, but daily. Without fail!

• Hippocampal Formation provides cognitive map of a familiar environment which can be used to identify animal’s current location. (see his first slides)

• Hippocampus - Seahorse like structure. Almost same in every animal’s.

• place cells - contain only the location, not anything else! because when it comes to the point where a particular place cell becomes active, it is independent of velocity or direction in which it is moving. There are no other stimulus in close cylindrical box.

• Different place cells become active in different locations.

• Place [PLACE CELLS] , direction [DIRECTION CELLS], distance [GRID_CELLS] are three important things in order to navigate - ie. these info must be stored in brain cells. So, there are PLACE_CELLS, where are other cells?

• Moving with marking cues - helps a lot in navigation. place cells are fired as a response to those cues. Hence, better navigation.

• Cues on the side are important rather than the ones on base. (why?) Even when the mouse does not PHYSICALLY run down the environment (passive mode in navigation ie. shown as if he’s navigating but with no physical movement. It was done using virtual reality software developed for mice as talked in the video), the cues still fire ie. 25% Field (aka scene) still maintained in passive mode.

Path Integration

The cells still fire in the right place if s/he has cues in the starting and the lights are turned off, ie., s/he has hints (~50% of the time) to her/his initial position based on which, using her/his own initial frame, s/he can calculate her/his navigational path (ie. place cells fire in same place again!) BUT, if s/he had no cues in the starting, s/he has very little idea where s/he is and cells do not fire at place where the cues were placed earlier.

So, how many maximum numbers of environments can we distinguish?

There are different [quantised] levels of representations (QLoR) of the place (from local ie. a room, to global ie. city or state). In different parts of hippocampus, these QLoRs are stored and communicated. No one knows how - the communication - hence calculating the max. number of environments is non-trivial.

This is coherent with the idea of ‘invariant representation’ mentioned in ‘On Intelligence by Jeff Hawkins’ which mentions the 6 layered neocortex and each level representing different [level] information in order to create an invariant representation that can further be used for other tasks (?) // TODO : re-read ‘On Intelligence’ and substantiate [think] more.

Problem? : How can we tell the neighboring location from this location? Oops. No one knows, yet!

How is the structure of hippocampus different from other areas of the brain and how does it, if at all, explain location of place cells?

Classic, function-structure relationship question for brain parts! Hippocampus is a part of primitive cortex. Hippocampus does not have all the layers of cells that are present in neo cortex. Very simple structure.

If seen from computational neuroscience perspective, it is continuous attractive network - because you get huge connectivity between principal cell types. Cells talk to each other at lot and to a lot of cells within. Associated network because each fibre that comes in runs along many many cells ie. one incomer fibre provides information to a lot of cells in hippocampus at the same time. + Existence of first-past-the-post inhibitory networks (who fires first inhibits others from firing within certain radius)

How does gravity affect navigation? What if the experiments are done on tilted platform?

If you turn animal upside down, the signal is lost. Most of the work has been on 2D. There IS a gravity signal. But it is not studied very much.

What else I think?

• if narrative in spatial frame => hippocampus active during navigation. – how about the same in time? any CORRELATION?