Conceptual Knowledge Flashcards
Parallel Processing and distributed knowledge represent the
Connectionist Approach
connections between neurons vary in
strength
connection between neurons can be
facialiatory or inhibitory
facialiatory connections are given
positive weight
Connections between synapses can change through
synaptic consolidation (fire together wire together)
The connectionist approach involves:
Creating computer models for representing cognitive processes
Parallel distributed processing
Knowledge represented in the distributed activity of many units
Weights determine at each connection how strongly an incoming signal will activate the next unit
What are the 3 types of units in the simple connectionist model?
Output
Input
Hidden
The activity of a unit is the sum of
The activation weights of each unit that are input to it.
Summation Effect
Learning in a Connectionist Model
Stim input to network
Stim info propagates thru network
Accurate Feedback
Error signal (diff between activity and correct activity)
Back prop- error signal back to start
Weights change to match output to correct signal
repeat til error signal is zero
Disruption of performance as system is damaged
Graceful Degradation
SPAUN is
Semantic Pointer Architecture Unified Network
SPAUN has which neural regions repped?
Pre-Frontal Cortex- dorsal lateral, ventral lateral, OFC
Visual 1-3 and Inferior temporal cortex
Thalamus
Posterior Parietal cortex
SPAUN can read and process
numbers, letters
SPAUN shows which two effects related to working memory?
Recency and Primacy
Limits of Spaun
- Simulates only a part of the full brain
- Can’t learn new tasks
- Can’t process more than a few numbers and a few symbols
- Computationally slow