Brandon - Spatial Learning Flashcards

Question

Which cells are considered to act as our internal compass? What are an important feature of them?

Answer 1

Head direction cells They are hard-coded, not experience based → pups had them in P11 pups (pups open their eyes for the first time at P8-9) *Recorded in the Pre-subiculum and para-subiculum of the pups

Answer 2

1. DTN system has Angular Velocity cells (AHV) → inhibitory signal onto LMN 2. Lateral Mammillary nucleus (LMN) = AHV cells and HD cells (head direction) LMN → ADN (Anterodorsal thalamic nucleus) → Pre/post-Subiculum → Entorhinal cortex → Hippocampus *ADN is the first nucleus to get vestibular input

Answer 3

DTN (Dorsal tegmental nucleus in vestibular system → AHV cell) LMN (Lateral mammillary nucleus → AHV and HD cells) Evidence: 1. Lesion in DTN imparis ADN HD cells 2. Lesion of LMN impairs ADN HD cells 3. Lesion of ADN impairs postsubiculum HD cells 4. Lesion of postsubiculum does NOT imapir ADN HD cells *Hierarchy from vestibular → thalamus → cortex/hippocampus (require input for the lower areas)

Answer 4

1. Dorsal tegmental nucleus → angular velocity cells 2. Lateral mammilary nucleus → mostly head direction cells (maybe a bit of Angular velocity cells) 3. Anterodorsal thalamic nucleus → 75% head direction cells 4. Post/pre-subiculum → head direction cells 5. Entorhinal cortex → head direction cells, grid cells, place cells 6. Hippocampus → place cells

Answer 5

LMN → broad tuning curve (~180˚) ADN (Anterodorsal thalamic nucleus) → more narrow tuning curve + anticipatory firing Postsubiculum → even more narrow (no anticipatory, because the head direction has caught up)

Answer 6

They exhibit "anticipatory" firing The peak is a bit shifted dependent on wether the mous is turning CW or CCW anticipating the future head direction → postsubiculum neurons, when they get the input, they correspond exactly to the direction of the head

Answer 7

They depend on vestibular inputs Lesions hair cells from the vestibular system with Sodium arsanilate (into inner-ear) → ne more response/firing form the HD cell of ADN

Answer 8

1. Mouse starts in a little refuge and is let free in a completely dark room 2. It encounters a large food pellet - Natural instinct is to bring the large food pellet back/run with it 3. The mouse runs back to the refuge → slight error in spatial navigation, the mouse gets close, but not straight to the refuge Compair this error with the shift in the head direction cells tuning - In the dark, HD cell tend to shift a bit creating an error in comparison to the environment They did show that the error in returning angle to the refuge correlated to the shift in head direction cells

Answer 9

Put mouse in a box where can climb everywhere. Start on 1 side and want to get to food on the other side, but they have to climb by the ceiling to access it because the direct floor path is blocked. When climb the walls → head direction cells fire as if it was a continuation of the floor On the ceiling → head direction cells stop firing *Not great study model because rats don't live in 3D

Answer 10

Used wireless recordings on bats. 2 hypotheses: - Spehrical coordinates - Toroidal coordinates (actual case)

Answer 11

Toroidal coordinates: - Azimuth cells → only care about the azimuth no matter the pitch - Pitch cells → only care about the pitch no matter the azimuth - Azimuth x Pitch cells

Answer 12

They replay different direction of the head (sets of neurons related to different directions are activated → seen by decoder) Slow wave Sleep → lots of head direction motion REM → slow rotation in head direction (according to the sets of neurons firing)

Answer 13

Put rats on a spherical stair case → no notion of Z-axis, the same X-Y place cells fires at every turn of the plateform at the same location Bats have X-Y-Z place cells

Answer 14

An attractor state is a set of neurons that fire corresponding to a specific object/ location Ex: in the duck/rabbit illusion, we either fall into the duct attractor state or the rabbit attractor state → takes energy to switch from one to the other

Answer 15

Pattern completion = when stuck into 1 attractor state, even when noise is added/input changes a bit, the output stays the same (see the same thing) Pattern separation = When the input changes too much, the output also changes → change attractor state

Answer 16

*Hippocampal CA1 place cells Global remapping: A place cell fires in very different location in 2 different rooms/fires in one room, but does not fire in the other room - For very different rooms *Change attractor state → associated with pattern separation Rate remapping: A place cell fires in the same location in 2 rooms, but at different rates - For same room, but change wall color for example *Stay in the same attractor state (same subset of cells firing) → associated with pattern completion

Answer 17

They do so by changing their firing rates T-mase where the mouse goes right, then left, then right, then left and gets food at start point if he does it good → Place cells fire more/less depending on if the animal has to go right or left (shows memory form 30 seconds ago)

Answer 18

Yes CA1 place field formation is not dependent on Schaffer collaterals coming from CA3

Answer 19

Grid cells from the Medial Entorhinal Cortex, through the Layer III perforant path *Mostly dorsal MEC (ventral MEC have weaker spatial tuning)

Answer 20

Small envrionment → place fields take a bigger portion of the envrionment/appear bigger Larger envrionment → more firing spots, can better the see the grid pattern, more regular *Due to boundary effects

Answer 21

Grid pattern: - Hexagonal grid of fields - From the center → ~60˚ shifted lines of firing spots → equilateral triangles - Gird cells fire in complete darkness (following the internal compass, don't require visual input) - All grid cells fire in all environments - Immediately formed in new environments and evolve/get more complete as you navigate it

Answer 22

Can't really record from the MEC in humans → measured Blood flow/glucose around the grid cells? As grid cells have shifted lines of firing spots every 60˚, make the person move in a virtual environment and shift direction a bit → every 60˚ they an increase the BOLD activity

Answer 23

Entorhinal grid cell realign (shift in the place fields) → global remapping

Answer 24

Dorsal → more numerous smaller place fields *Progressive increase Ventral → fewer larger place fields, more spaced out *Observed by making mice run on a linear track and recording form different areas Hippocampus: Dorsal → more narrow place field Ventral → larger place field (even up to 10m diameter)

Answer 25

Grid cells encode each hallway very similarly as if they where the same envrionment (1 out of 2 because the go north, come back south, then north, etc.) → Fragmentation of grid cells *Hippocampal place cells do the same thing

Answer 26

Lesion in the medial septum → no input onto Medial Entorhinal Cortex (can still fire, but no grid cell spatial firing pattern) *Eliminates theta oscillations throughout the medial temporal lobe Then they tested CA1 place cells in a familiar and novel envrionement: → CA1 place cells were suprisingly able to develop new place fields in a novel room without septal inputs and without grid cells → They also conserved the place fields in familiar rooms Hypothesis: different cell types in the medial entorhinal cortex can drive formation of place fields, not only grid cells

Answer 27

1. Grid cells → no directional tuning 2. Head direction cells 3. Conjunctive cells → grid patterns + head direction tuned 4. Boundary vector cells → create large field that spans 1 axis of the envrionment and codes for distances from the boundaries of the envrionment 5. Speed cells → firing rate of thses cells is linearly correlated with runing speed

Answer 28

Boundary vector cells 2 or more boundary vector cells → input together → Place cells to fire in a specific location *Not affected by medial septum inactivation/lesion ~20% of MEC = grid cells ~2% of MEC = boundary vector cells

Answer 29

Time cells Experiment: 1. Have a rat run for 15secs on a treadmill 2. Door opens 3. Alternating T-maze task 4. Repeat Lots of cells firing in the 1st second, but different cells firing after 2, 3, 4, 5... seconds Grid cells can code for distance and time of the treadmill task, some will fire only at 1 place/time, some will fire every 3 seconds, etc.

Answer 30

- Largest oscillations in the mammalian brain - 6-10 Hz (relatively slow) oscillations in local field potential in the hippocampus and related structures - Observed during exploratory and attentive behaviours - Seen in rats, mice, gerbils, rabbits, bats, pigeons, monkets, humans - Reflect highly and precise spike timing throughout the hippocampal structures *every 125ms ~ 8Hz

Answer 31

It is a caused single unit acceleration of the spiking activity of a neurons when the animal enters its place field Seen when animals ran on a linear track: - At the start of the place field, firing at the peak of theta oscillation - Slowly shifts at the neurons has a slight increase in firing rate (~10Hz instead of 8Hz for ex) - At the middle point of the place field, the neuron fires in the bottom of theta local field oscillation - At the end of place field (location), firing has shifted back to the peak of the theta oscillation - When exit the place field, the neuron's firing rate falls back into the theta oscillation rhythm it was following before the animal entered it place field *Correlation (negative slope) in the delay between peak or theta oscillation and firing CA1 place cell and where the animal is in the place field *When walking, shift slower than when running as advance slower in the the place field

Answer 32

*Occurs when the animal is in exploratory, attentive behaviour - In entorhinal layer II Grid cells - CA1 place cells - NOT entorhinal layer III grid cells → phase locked, meaning they only fire following the theta oscillations regardless of position on the track

Answer 33

Phase precession is the result of the neuron oscillating faster than the extracellularly recorded theta oscillation Did so by comparing extracellular field and intracellular recordings: - Fix head of mouse so it can't move (for intracellular recording) - Body can still move and is on a ball - When the mouse runs on the ball, it navigates in this virutal world in front of it

Answer 34

Precise temporally structured cell assemblies Different place cells which have neighbouring place fields are just a bit delayed creating an order of their sequence of activation - When the animal enters a new place field, that cell starts firing with theta oscillations, then under goes phase precession and comes back to theta rhythm at the end of the place field Halfway through the track, the hippocampus has a redout of the the place fields from the start of the track to the end of the track within 125ms (1 theta oscilation) as they are all closely delayed in sequence of activation

Answer 35

Sharpe-wave ripples are events that occur in non-theta states, when the animal is still (when there are no theta oscillations) These Sharp-wave ripple events propagate throughout the hippocampal formation (CA1, CA3, Subiculum, EC, Parasubiculum, etc.) - Activation of pyramidal cells *Occurs during slow wave sleep and daydreaming/resting, conserved across species (not REM sleep or exploration/attention/running) Sharp-wave = 1 slow, but very large deflection of 1-50Hz / ~2mV Ripple = many quick oscillations on top of the sharp-wave (1Hz - 10kHz, ~ 0.5mV) → Sharp-wave ripples trigger activity throughout ALL of cortex (not subcortical areas, done by FMRI bold signal recording)

Answer 36

Before → Shortwave ripple event activating all place cells that will be activated during the run in the sequence they will be activated (Pre-play) During the run → Place cells are activated and undergo phase-precession in order of place fields After the run → Sharpwave ripple event activating the place cells, but in reverse order (Replay backwards) *During sharpwave ripples, there is decoding of the position of the animal (place cells fire as if the animal was in a specific location, but the animal is resting not at that location)

Answer 37

Setup: 1. Put a mouse in the middle with 6 linear tracks (forming a star) 2. Put food baits at the end of 3 of these tracks By the end of the day, the animal knows which tracks have the food 3. Allow the animal to rest for 1 hour outside the task envrionment During this hour: - Control group → After each sharpwave ripple event, silence neuronal activity for a short period - Experimental group → At the start of each sharpwave ripple event, silence neuronal activity for a short period (no replaying/ripple events for memory consolidation) 4. Redo the task → experimentl group has worse results than control group who did have consolidation through shortwave ripple events (replaying) To silence neuronal activity, they stimulated the commissure pathway connecting right CA3 to left CA3 → causes almost all hippocampal neurons to fire at once → causes a general synchronized refractory period of 200-300ms

Answer 38

- 30-100 Hz oscillations in local field potential in hippocampus and related structures - Co-occur with hippocampal theta oscillations - Thought to function to coordinate activity between distant structures - Seen in rats, mice, gerbils, rabbits, bats, pigeons, monkeys, humans Experiment showed that hippocampal gamma oscillations increase prior to choice point on an alternation taks: - Circular alternating T-maze with delayed section (treadmill) - Gamma oscillations were weak when animal was coming back to the delay zone - Increased in power as the animal was approaching the point of decision of the T-maze → where it had to retreive information from memory (what did it do last time?)

Answer 39

Slow (~40Hz) gamma reflects CA3-CA1 synchrony Fast (~90 Hz) gamma reflects MEC-CA1 synchrony *All regular within the theta oscillations

Answer 40

Setup: Animals exposed to odor cube → if given odor A, have to go to food pellet A, if given odor B, have to go to food pellet B - LEC is though to be responsible for odor processing - Animals learned the task for 5 days There was Gamma coherence between LEC and distal CA1 supporting the associational learning: - LEC and dCA1 are connected (need to be connected to an area to have coherence, MEC is connected to proximalCA1) - At day 1, weak coherence between LEC and dCA1 - At day 5, much stronger coherence between LEC and dCA1 + much better performance at the task - In the failed trials at day 5, decreased coherence between LEC and dCA1 for that specific trial was observed (shows coherence is required)

Answer 41

1. Sensory input stimulates 1 neuron af attractor A 2. Neuron A spreads the signal to the other neurons of attractor A (excitatory connections) 3. Neurons from attractor A send inhibitory signal to all other attractors (neurons) → Feedforward inhibition Compeition / selection of attractor networks

Answer 42

Continuous attractors → head direction *Signal is generated at the levels of Dorsal tegmental nucleus (DTN) → Lateral Mamillary nucleus (LMN) Attractor model requires recurrent excitation + lateral inhibition: - Connections are organized in a ring - When facing north → sensory input activates the neuron "at the north of the ring" - This neurons sends excitatory input to neighbouring neurons → recurrent excitation - This neuron sends inhibitory input to neurons farther in the ring This allows maintenance of the signal (by recurrence) as neurons operate on a much faster time scale as perception

Answer 43

Drosophila had a ring of neurons coding for head direction Experimental setup → fixed head, free body on a spining ball, the drosophila looks at a screen with a LED - Look at deconded position of animal (ring signal) vs LED

Answer 44

In grid cells, every neuron, when activated, will send excitatory signal to its neighbouring neurons and inhibitory signal to neurons farther in the field → Easy to produce place cells form this system For grid cells, neurons that are connected by excitatory input → theoretically close if you fold the space field (neurons of the EC and their place fields) into a doughnut shape → close neurons in the doughnut form excitatory connections By using the attractor model, we can visualize a "sheet" of neurons as if it was folded into a donut shape. Then, if the animal runs on the surface of the donut (in circle), we can see how grid cells form this triangular pattern. The only parts of the donut that the animal runs along make up activations that form the grid cell spots you see when you unfold the donut. *Explains the grid cell pattern is the space

Answer 45

Grid cells Oscillatory interference model → generates time cell firing

Answer 46

Model → Oscillatory interference model 1. Intrinsic frequency (from dendrites/sensory input, faster~9Hz) 2. Network frequency (from sooma ~ 8Hz → theta oscillation rhytm) Sum up → When out of phase ~ 0 amplitude, When in phase ~ Large amplitude - Set a threshold for spiking → Spiking pattern of regular bursts based on the sum of these frequencies *Requires both rhythms to be constant By inactivating the medial septum (supporting the network frequency of hippocampal theta oscillation) → no summing → loss of the timing properties Thiming cell need theta oscillations

Answer 47

Theta oscillation (supported by medial septum) → for summing with the intrinsic frequency

Answer 48

Sum up these different frequencies: 1. Baseline theta (sooma) 2. Input from different head direction cells trough dendrites at different frequencies (can have many different frequencies) When all signal are enough in phase for the sum to surpass the threshold → spiking This spiking pattern/sufficient sum will occur in different locations in the RF creating a grid pattern

Answer 49

*From the oscillatory interference model There is a gradual increase of grid scale along dorsal/ventral axis → Correlates with an decrease in intrinsic frequency of MEC neurons along dorsal/ventral axis Dorsal → smaller grid scale → higher intrinsic frequency → more often, but shorter period over spiking threshold Ventral → larger grid scale → lower intrinsic frequency → longer periods over the threshold, but less often

Answer 50

Encoding and Retrieval: Theta rhythm modulation of LTP for encoding dynamic EC → LTP onto CA1, but also prefer to fire at the peak of theta wave CA3 → LTD onto CA1, but also prefer to fire at the trough of theta wave

Answer 51

Theta wave dictate encoding vs retrieval patterns and separates them in time (different phases of the wave). Encoding: - Peak of theta wave - Strong EC input (→ CA1 and CA3), weak CA3 input onto CA1 - Strong LTP Retrieval: - Through of theta wave - Weak EC input, Strong CA3 input (→ CA3 (recurrent) and CA1) - LTD/depotentiation

Answer 52

When encoding of a new memory and retrieving of an old memory occur at the same time in the hippocampus → CA1 *They normally should occur at different phases of theta oscillations

Answer 53

ACh highest when walking/active and lowest in Slow wave sleep (medium when resting awake) Cholinergic presynaptic inhibition of glutamate release in cortical structures (hippocampus, olfactory, EC) High ACh: Strong input from EC → CA1 (encoding), but inhibited feedback within the hippocampus (retrieval) *Neocortex → Hippocampus Low ACh: Weaker input from EC → CA1 (encoding), but strong feedback from CA3 (retrieval/consolidation) Low ACh → increase sharp wave ripple → increase schaffer collaterals + CA3 recurrents *Hippocampus → Neocortex

Answer 54

It block sharp wave ripples *Medial Septum stimulation → responsible for cholinergic input onto the hippocampus → blocks sharpwave ripples - Correlates with less consolidation and more encoding - When low ACh, more sharp wave ripples and more consolidation/retrieval

Answer 55

1. Local effect of ACh in the hippocapus blocking reafferent inputs during active state/encoding 2. Theta oscillations vs Sharp wave ripples