Lecture 20 Flashcards
The basal ganglia
Dopamine signaling here regulates motivational processes and reinforcement
In the absence of this - no purposeful movement
Muscles work fine, just don’t do anything other than instinctive movements
Will starve next to food. Wont eat if you put the food in their mouths. Aimlessly swim
The hippocampus basics
Needed for explicit LTM
No functional hipp cannot form semantic or episodic memories
STM is fine and so is older semantic info. Focus on hear and now, no dwelling on past of predicting future.
T maze experiments
Cross shaped maze. Rats trained until they can do it 90% accurately. Food in one branch.
If rat learns by following directions - response learning strategy
If rat learns via landmarks, place learning strategy
What is initially learned is place
This means it is an explicit spatial memory
With a lot of time, stimulus-response memory dominates and the rat learns by response learning strategy
Spatial and response learning in humans
In fMRI humans navigate through VR map that allowed spatial clues or series of turns. Half learned space, half response. If hipp larger and more active, spatial. If not and caudate nucleus more active, response.
If the task was switched so that only one strategy worked, the ones with larger and more active systems that did not work for this new maze ie hippocampus for the maze that only works with response strategies: they make more errors.
Morris Water Maze
Big tank with hidden platform
Control - rapidly learned where platform was, subsequent trials show very fast reaching of platform.
Conscious explicit learning of platforms location.
Lesion hippocampus - after 12 trails, as good as control. Had learned implicitly the route from the same start point to the platform.
BUT if the rat is placed at a random start point every time, control adapts no worries using explicit clues. The response learning strategy doesn’t work and the rats with lesions in their hippocampi cannot ever learn a way and so get no faster
Dopamine signaling lesion = aimlessly swim
Configural learning
People perceive configurations of sensory stimuli (ie whole face) rather than specific, isolated incidents.
It is thought that before we can commit sensory details to LTM we must develop a map/configurational rules for the structure of the sensory information.
Spatial learning
As someone walks around a grid, they not only learn the grid, they learn where they are relative to where they have been. The person builds a map. If you know the structure of a space, you only have to visit each area once and you can predict what you would see if you took a different route though the structure. Without this, you would need to learn every single way through that structure.
This also happens with social structures. If someone is born to a family, you immediately know their relations to everyone else via this map/scheme.
The main inputs to the hippocampus
Come from the entorhinal cortex on the temporal lobe.
The medial entorhinal cortex (MEC) is thought to provide learned structural info
The lateral entorhinal cortex (LEC) is thought to provide perceived sensory info
The hippocampus (HC) combines it, associating sensory information with place (position on a map or an abstract theoretical structure).
Neurons of the MEC (integration)
The spiking activity of different populations of the MEC correlates with
1) the speed of an animal as it walks around
2) the direction an animal faces
These two variables are enough to make a map by integration: you can know where you are relative to where you started and hence make the quickest route home known.
Map making
Animals sense of direction initially depends on path integration. After a while the information accrued during integration promotes map formation and map-based navigation.
Neurons of the MEC (Grid cells)
Grid cells in the MEC fire as animals walk. They seem to represent a hexagonal grid and fire when the animal is in the same position in the same room, reliably. There are thousands of grid cells and each neighboring grid cells have similarly sized grids that are rotated.
They are defined by the space between hot spots and where they are relative to others.
Summed, they allow precise location of the animal
When an animal walks into a different space, the grid of every grid cell rotates. This is called grid cell realignment.
As you move dorsal to ventral in the MEC, the hot spots of the grid cells separate. There are 5-10 groups each a bit farther apart.
This allows hierarchal organization of the encoded structure.
Allows immense combinational power, new spaces can be mapped without interfering with old spaces.
Neurons of the MEC (Border Cells)
These fire when an animal is located near a wall in a particular direction.
Neurons of the MEC (Object Vector Cells)
Fire whenever an animal is a certain distance and direction from an object
Place cells in the Hippocampus
Within the hipp, many neurons only fire when the animal is in a specific position in a room.
These can even include what direction one is moving in, easily shown in narrow passageways
These place cell firing patterns are consistent from day to day. But the position encoded by the cell in one room does not account for the position it encodes in another.
Accordingly, THESE CELLS PROBABLY REFLECT THE CONOF THESE 2 STREAMS: WHERE AND WHAT AND PROBABLY WHEN
Grid cells and place cells
A recording of one grid cell while an animal explores 2 rooms will differ as the animal undergoes grid cell realignment when moving between the two rooms (the rotation changes, not the distance).
Similarly the place cell changes its pattern as the place cell firing is dependent on a particularly grid cell alignment.
The same corridor can provoke different patterns of place cell firing depending on whether an animal attempts to turn right or left at the end.
