Task 5 Flashcards
(52 cards)
What is Hebb’s Law?
When a neuron stimulates another neuron while the receiving neuron is actively firing, the connection between them strengthens.
What are some limitations of Hebb’s Law?
It does not specify how much connections should strengthen.
It lacks a mechanism for reducing connection strength.
It does not define the conditions under which learning should occur.
Why is Hebbian learning difficult to implement in computer models?
Because it allows connection strength to increase indefinitely, leading to computational instability.
What is Neo-Hebbian learning?
A refined mathematical formulation of Hebb’s Law using differential equations to describe neuron activity and weight changes over time.
What is a neurode?
An artificial neuron with multiple input signals and one output signal.
What are instar and outstar neurodes?
Instar – A neurode that receives input signals.
Outstar – A neurode that sends output signals to multiple receivers.
How does instar-outstar learning relate to neural networks?
Each neurode is both an instar (receiving input) and an outstar (sending output), allowing complex learning and adaptation.
What is differential Hebbian learning?
A variation of Hebb’s Law where connection strength changes based on the difference in neuron activation rather than absolute activation.
How does differential Hebbian learning differ from simple Hebbian learning?
If neuron activity stays constant, no learning occurs.
Weight changes can be positive or negative, allowing for adaptive learning.
Why is differential Hebbian learning a better biological model?
Because biological neurons must allow for both strengthening and weakening of connections.
What is Drive-Reinforcement Theory (DRT)?
A learning model based on differential Hebbian learning that includes time-dependent memory formation, making it closer to classical conditioning.
How does DRT improve upon traditional Hebbian learning?
It incorporates time-based learning, not just present-moment stimulus.
It explains S-shaped learning curves, where learning starts slow, speeds up, then slows again.
It considers recent history of stimuli, rather than only current inputs.
What happens when the hippocampus is damaged?
New episodic memories cannot be formed.
Old memories remain intact.
Procedural memory (skills) is unaffected.
How does the hippocampus receive information?
It gets input from the parahippocampal gyrus and entorhinal cortex, which collect data from multiple sensory areas.
What are the three main stages of information processing in the hippocampus?
Dentate gyrus (DG) – Filters and sparsely encodes inputs.
CA3 region – Performs autoassociation for memory recall.
CA1 region – Transmits processed information back to the cortex
What is the perforant path?
The main input pathway connecting the neocortex to the hippocampus, providing sensory and contextual information.
What is the role of the dentate gyrus (DG)?
Creates sparse representations of input signals.
Reduces overlap between similar memories.
Improves the storage capacity of the hippocampus.
What is the role of CA3 in memory?
Acts as an autoassociative memory network.
Uses recurrent connections to strengthen memory recall.
Allows cued recall (retrieving a full memory from partial input).
How do recurrent connections in CA3 aid memory?
They allow CA3 neurons to reinforce each other, improving long-term recall and pattern completion.
What is the role of CA1?
Transmits processed memory data back to the cortex.
Further refines memory recall before sending it to storage.
How does sparse encoding improve memory?
By ensuring that similar events are stored with minimal overlap, reducing confusion during recall.
What happens when recurrent CA3 connections are removed?
Memory recall worsens.
CA3 activation patterns become more random.
Later memory stages (CA1, entorhinal cortex) also perform worse.
How was the hippocampal memory model tested?
100 random memory patterns were stored using Hebbian learning.
Fragments of the patterns were later used as retrieval cues.
Successful recall occurred when the retrieved memory correlated with the original input.
How does episodic memory formation differ from procedural memory formation?
Episodic memory – Forms quickly, stores individual events separately.
Procedural memory – Forms gradually, blends multiple experiences into a knowledge base.
