11. Systems Electrophysiology of Spatial Navigation, Learning and Memory

1. declarative / explicit

2. non-declarative/ implicit

episodic (events)

semantic (facts)

• skills & habits

• priming

• motor reflex learning

• emotional responses

• encoding (processing of information into the memory system)

• storage ( retention of encoded material over time)

• retrieval ( process of getting the information out of memory stage)

• The hippocampus stores patterns immediately and efficiently as classificatory units.

• It does this without further analysis.

• It serves as an intermediate memory, temporarily holding information.

• Over time, the neocortex extracts important features and consolidates them, reducing redundancy.

• This is a slow transfer process from hippocampus to neocortex for long-term storage.

Long-term potentiation (LTP) is a persistent strengthening of synapses based on recent patterns of activity.

• Mnemonic techniques are memory strategies that help encode and recall information more effectively.

• Example: Method of loci (Memory Palace), which places items along an imagined spatial route.

• These techniques activate brain areas involved in spatial navigation, especially the hippocampus.

• The brain reuses spatial memory systems to structure abstract information.

• This link explains why spatial imagery is so effective for long-term memory.

• Path integration: Self-motion-based navigation (using movement cues) to track position from a starting point. →( Relies on grid cells in the entorhinal cortex.)

• Allocentric navigation: Map-like navigation using external landmarks. →( Relies on place cells in the hippocampus.)

• Both systems work together in real navigation.

• You track your movement (path integration), but correct errors using known places (landmarks = allocentric).

1. Egocentric: Relative to your body (e.g. “left of me”), updates as you move.

2. Allocentric: Relative to the environment (e.g. “north of table”), stays stable.

3. (Brain: Egocentric → parietal cortex, Allocentric → hippocampus.)

• NMDA receptors open only when: → Glutamate is present (presynaptic firing) and → the postsynaptic membrane is depolarized.

• This allows Ca²⁺ to enter, triggering LTP (Long-Term Potentiation).

• The NMDA receptor detects when both neurons fire together → = a “coincidence detector” that strengthens the synapse only if both sides are active at the same time.

• If the presynaptic spike fires before the postsynaptic spike (within ~20 ms): → LTP (synapse is strengthened).

• If the postsynaptic spike fires before the presynaptic spike: → LTD (synapse is weakened).

• STDP = Spike-Timing Dependent Plasticity. → Learning depends on precise temporal order.

1. Nernst equation: → Calculates the equilibrium potential for one ion. → Based on concentration difference and ion charge.

2. Goldman equation: → Calculates the real membrane potential using multiple ions. → Also includes membrane permeability for each ion.

• High-frequency (>500 Hz): → Spikes from single or multiple local neurons.

• Low-frequency (<300 Hz): → Local Field Potentials (LFPs) — reflect synchronous synaptic input from upstream projections.

• Uses a thin metal wire (20–50 μm) → small tip = high impedance.

• Typically records 1–2 neurons with high SNR, but can't clearly separate all spikes.

• Needs preamplification due to long cables.

• Enabled by small, robust Field Effect Transistors (FETs).

• Cannot easily distinguish spikes from many nearby neurons.

• Neurons lock to stimuli and are feature-selective.

• Show tuning curves (e.g. to frequency, orientation).

• Have clear topographic organization (e.g. somatotopy).

• No global synchrony or large-scale population dynamics observed.

• No direct locking to stimulus, but show spatial tuning.

• No topographic map like in sensory cortex.

• Show global synchronous oscillations (e.g. theta rhythm).

• Involved in complex population dynamics (e.g. sequences, replay).

• Neurons in the medial temporal lobe that respond to specific concepts (e.g. Halle Berry).

• Fire across different modalities: photo, name, drawing, imagination.

• Reflect abstract, high-level representation (not just visual features).

• Support sparse coding in memory-related areas like the hippocampus.

• A known current source in tissue creates a voltage field.

• Uses electrostatics with substitutions: → Charge → Current, → Permittivity → Conductivity (σ).

• Voltage at distance R decreases with distance and depends on conductivity.

• This is called the forward model: from current (I) to voltage (V).

on electrode position

A tetrode enables multi-unit recordings by capturing spike waveforms across four wires.This allows reliable identification of individual neurons in dense neural populations.

• Allow precise electrode size and spacing, reproducibly manufactured.

• Enable hundreds of electrodes on small, compact probes.

• Exact position of each contact in the brain is known.

• Support linear and 3D array designs.

• Record many neurons simultaneously with high spatial and temporal resolution. → Ideal for studying population activity and brain-wide dynamics.

• Place cells fire at specific locations.

• The animal’s current position is encoded by the combined firing pattern of many neurons.

• This population code changes dynamically as the animal moves.

• Location can be decoded from the neural ensemble’s activity.

• Attractor dynamics describe how neural systems settle into stable activity patterns.

• An attractor is a stable state of a neural network that the system "settles into."

• Used to model memory, decision states, and spatial coding.

• Types: point, line, ring, or continuous attractors.

• Important for explaining how the brain stores stable internal representations.

1. McNaughton: Head-direction cells provide the angular input needed to update spatial position during movement.

2. Yoder: Lesions in head-direction pathways impair path integration, even with intact speed signals. → Path integration fails without direction reference.

• Yes, grid cells in MEC can remap.

• Hafting et al. (2005) discovered grid cells with hexagonal firing.

• Later work (incl. Buzsáki) showed they shift, rotate, or scale depending on environment or task. → Grid cells provide a flexible spatial metric, not a fixed map.

• Found sudden transitions between hippocampal maps when rats explored intermediate environments.

• Showed that hippocampal population activity reflects discrete attractor states.

• Supports idea that the hippocampus uses attractor dynamics to stabilize spatial memories.

You can decode an animal’s location by recording neural activity using tetrodes, which capture the firing of spatially tuned neurons like place and grid cells, and applying methods like population vector or Bayesian decoding to estimate the animal’s position from the activity pattern.

Using tetrodes, you record the firing of spatial neurons over time, and by comparing this activity to their known spatial tuning (via methods like Bayesian decoding), you can accurately reconstruct the animal’s position at each moment.

• A tetrode records electrical signals from multiple neurons simultaneously, using four closely spaced electrodes.

• This allows spike sorting, i.e., identifying which neuron fired based on the unique spike shapes across channels.

• From this, you get time-resolved firing patterns of individual neurons (e.g. place cells).