5-Plasticity Development Neural Circuits

• Activity-independent

  —> genetically and molecularly driven processes, that dont require neural activity or sensory input

  e.g (=basic structures of nervous system)

  • neuron generation/differentiation

  • axon guidance to target areas

  • initial formation of synapses

• Activity-dependent

  —> depend on neural activity and interactions with environment -> shaping brains functionality

  e.g

  • Spontaneous activity

  • Sensory-evoked activity: stimuli from environment strenghten/weaken synaoses through synaptic plasticity

• neural connectivity = basis of clustered networks

• synaptic plasticity is influenced by neural dynamics (e.g spiking activity) —> modifies strenght over time

—> network can exhibit attractor activity -> stable patterns of neuronal firing

attracors: essential for memory formation, decision-making

‘neurons that fire together - wire together’

when axon A repeatedly excites cell B, growth process increases A’s efficiency in firing B

-> underlies Long-Term-Potentiation (LTP) -> correlated activity strengthens synaps

foundational for learning and memory

Plasticity in feedforward networks depends on:

1. activity of presynaptic neurons - input, ρ(t)

2. activity of postsynaptic neurons- output ν(t)

3. synaptic weight changes: wj = F(pj(t), v(t))

not really, it has fixed points that represent possible outcomes

but stability requires sth like hard bounds on synaptic weights to ensure meaningful learning

Hebbian = strenghtens connections based on correlated activity

Covariance = adjusts weights based on covariance of pre- and postsynaptic activity

= Spike Timing Dependent Plasticity

biological mechanism of synaptic plasticity where relative timing difference (Δt) btw pre- and postsynaptic spikes determines:

• direction

• magnitude of synaptic strength changes

• Examination of Synaptic Efficiency

  strength tested at regular intervals by stimualting presynaptic neuron (with voltage-clamp at low frequency 0.03-0.06 Hz)

• Induction of Synaptic Changes

  syn. plasticity induced using repetitive stimulation

  60 pulses at 1 Hz to presynaptic neuron (both pre and post in current-clamp mode to allow spiking)

relative timing btw pre and postsynaptic spikes determine whether strength increases / decreases:

• Δt > 0: post after pre

  —> Long-Term Potentiation (LTP) - strengthening synapse

• Δt < 0: pre after post

  —> Long-Term Depression (LTD) - weakening synapse

• Relative Timing

  • Positive timing (Δt>0): Strengthens the connection (LTP)

  • Negative timing (Δt<0): Weakens the connection (LTD)

• Neuromodulation (neurotransmitter concentration)

• Firing Patterns (burstiness of neural activity)

Phenomenological (What?)

• use pre- and postsynaptic spike times or firing rates to calculate changes in syn. strength

• mathematical models that observe relationships btw acitvity and syn. changes

Biophysical (How?)

• resolve biological machinery + processes involved in syn. plasticity induction

• highly varying levels of biological detail

Triplet STDP: can explain data where pairs of spikes are repeated at different frequences —> more complex patterns

Pairwise STDP: only timing of two spikes

BTDP focuses on synaptic changes driven by burst of spikes

timing of burst over individual spikes

• Additive weight change

  syn. weight adjusted independently to current weight magnitude

• Multiplacative weight change

  weight changes are scaled by current weight magnitude

• Synaptic Bounds

  weight updates constrained by predefined bounds, ensuring physical and computational stability

  -> bounds prevent weight from becoming unrealistic large or negative