L19 - Localization of learning and memory

Question 1

How is the hippocampus linked to spatial navigation in rats?

Answer 1

• Helps rats to understand the space around them
• Study: Create water bath with platform, rats will swim around until they find platform to stand on. Rat’s goal is to find the platform (which is not visible). Has to understand where the platform is based on extra maze cues.
  
  • Even when move maze, can understand where the maze is
  • Normal rat = can swim directly to the platform that it can’t see
  • Hippocampal lesion = Cannot work out where the platform is in relation to the room, just doesn’t know where to find something that it can’t see (but can find it when flag or something put to indicate platform)

Question 2

Where might we find changes in strength of synaptic connections underlying learning? Are these changes local or global?

Answer 2

• In keeping with localisation of function in brain, different types of learning appear to be localised in different parts of brain - e.g. local to area for language
• No specific areas though for learning, it depends on what is being learnt
• So global in the sense that lots of areas in the brain participate in learning (e.g. LTP found in spinal cord), all of sensory nervous system has capacity to learn
• Difference between learning in adulthood and learning in childhood

Question 3

What do we know about the hippocampus in humans when we are learning about spatial maps?

Answer 3

• Brain imaging studies in spatial navigation in humans
  
  • Activation of hippocampus while people navigate through a virtual town on a computer

Question 4

What are hippocampal place cells?

Answer 4

• Neurons that are active (fire) when rat is in specific environment
  
  • Allocentric receptive field → Organisational structure of space defined by relations among objects rather than within reference to observer which is the case in the receptive fields in the visual cortex
  • Don’t need to be in the particular place to represent the space, can be activated by imagining
  • e.g. Rat below, place cells fire when in particular part of arena. Consistent finding

Question 5

What inputs provide the information to place cells?

Answer 5

• Sensitive to visual input
  
  • Place cells continue firing in the dark and congenitally blind rats have place cells
  • Olfactory and tactile (whisker) inputs also influence spatial pattern - so are multimodal inputs
  • In example below
    
    • Place area is a white strip placed in the arena by experimenters. It moves as the white strip moves

Question 6

What is the location specificity of place cells?

Answer 6

• A single place cell is stable and specific within a particular environment, but can be active in >1 environment (as otherwise would require a cell for every location)
  
  • Individual cells do not uniquely code for a locations, pattern of activity across multiple cells does. Multiple cells will respond.
• Place cells tend to be non-directional (active in a location regardless of rat’s direction).
• Place cells form a “cognitive map” of environment.
  
  • Activity of cells can predict animal’s navigation through maze, and even predict its errors
  • Can see how place cells correspond to the mistake it makes in behaviour - can see place cell for where the food should be in the wrong area as that is where the rat thinks the food should be

Question 7

How do place cells learn about place?

Answer 7

• Place cells establish spatial pattern within few minutes of being introduced to novel environment and can maintain this pattern for days or even several months.
• Place rats into 2 similar enclosures. Place cell patterns were initially very similar, but after many exposures over several weeks, the patterns began to diverge.
  
  • Learning to discriminate between places.

Question 8

Where is the input to the hippocampus coming from?

Answer 8

• Inputs to hippocampus come from entorhinal cortex, where different types of cells provide building blocks for allocentric representation
• Grid cells in medial entorhinal cortex (MEC) that have spatial properties

Question 9

What are grid cells?

Answer 9

• Many neurons in MEC respond to rat’s position but follow a lattice or grid
• The grid pattern of a single cell can cover a large area (whole arena), and the cell will show same grid across different environments (Therefore, not about encoding particular places, but rather provide coordinate frame for any space.)
  
  • Grid cells retain exact grid layout despite changes in speed or direction of rat’s movement.
  • Pattern arises from intrinsic nature of network connections among cells in MEC
    
    • Short-range excitation between cells
    • Long-range inhibition
• Single grid cell provides ambiguous information about location (rat could be at any one of many location where activation strength is repeated)
  
  • But ambiguity can be resolved by combining across multiple grid cells that have different grid spacing and phase. Can also have different spatial frequencies
  • E.g. Below they have different sized areas that they fire in, different spatial frequencies so have different coordinates = help to distinguish place

Question 10

So what does the hippocampus do?

Answer 10

• Spatial processing may be a primary function of hippocampus, but spatial (and temporal) context is fundamentally important for many types of learning & memory (especially episodic)
• It is able to create allocentric representation of the world but can also do this within time which is why it is so important for episodic memory

Question 11

What roles does the cerebellum have in learning motor skills?

Answer 11

• Complex and intricate structure of Cb allows integration of sensory inputs for precise timing and sequencing of motor programmes (e.g. speaking requires control of muscles to be coordinated)
  
  • Cerebellum learns how to do these tasks

Question 12

What is the organisation of the Cerebellar cortex like?

Answer 12

• 3 key types of neuron
  
  • Purkinje cell = Output neuron
  • Climbing fibres = Inputs into cerebellum
  • Granule cells = Input into cerebellum
    
    • One purkinje cell will cover one climbing fibre = strong 1:1 relationship between them
• Granule cells pass through loads of purkinje cells
  
  • Any purkinje cells will have thousands of parallel fibres going through (parallel fibres = granule cells)

Question 13

How does cerebellar learning occur - what is the eyeblink reponse?

Answer 13

• Most extensively studied paradigm demonstrating learning in cerebellum is eyeblink conditioning in the rabbit (similar findings in humans).
  
  • Rabbit hears a noise (CS) followed by an airpuff to the eye (US). US normally elicits eyeblink.
  • After 30 or 40 trials, rabbit learns to blink in response to the noise
  • Their eyeblink is perfectly timed - don’t just hold it shut like humans do - perfectly timed to airpuff duration
  • Takes longer to learn this

Question 14

What is the machinery responsible for eyeblink conditioned response and how does it occur?

Answer 14

• Eyeblink UR (to airpuff) mediated by trigeminal nucleus and cranial motor nucleus (to nuclei responsible for the response)
• Eyeblink CR involves same nuclei as UR but more
  
  • Red nucleus in the brainstem
  • Pontine nuclei in brainstem
    
    • Sends signals to cerebellum
    • Electrical stimulation of pontine nuclei can serve CS when paired with airpuff of US
    • So Pontine nuclei is part of the pathway conveying CS input
• Inferior olive in brainstem
  
  • Receives input from trigeminal nucleus, output to cerebellum
  • Neurons respond to airpuff
  • Lesion prevents learning but no expression on previously learned CR (induces extinction)
  • Electrical stimulation elicits eye-blink and can serve as effective US (instead of airpuff) if paired with tone
  • Inferior olive involve in part of path of UR
• Cerebellum lesion prevents learning and blocks the expression of previously learned CR
• CS and US converge in cerebellum
  
  • Mossy fibres from pons synapse onto granule cells (carrying CS info)
  • Parallel fibres from granule cells form synapses with very many Purkinje cells: One PC receives synapses from 100,000 + parallel fibres
  • Info about the US (airpuff) come from the climbing fibres of the inferior olive
    
    • Strong exclusive communication between them: Each PC receives many (500+) synapses from one and only one climbing fibre
  • Potential for many CSs (or many different time signatures of CS) to be associated with US
    
    • Parallel fibres can change connections with ones being used as they are connected to so many
    • It is long term depression that is causing the learning here rather than long-term potentiation as purkinje cells send inhibitory signals to interpositius nucleus in cerebellum so with depression suppressing the signals from the PC then it means the interpostius nucleus can send signals
      
      • Cerebellum allows for the precise timing of movements

Damage to each of these structures eliminates eyeblink to CS but not to airpuff