Flashcards
Cognitive Psychology: Learning
Cards in this deck (22)
- 1What are the 2 types of learning?Associative and Non-associative (this one's found in all living organisms)
- 2What are the two sub-units of Non-associative learning?Habituation (decrease in response to repeated stimulus) and Sensitization (amplification of response to repeated stimulus)
- 3What is classical conditioning?Pavlov's dog example (bone + bell)
- 4What is operand conditioning?Modifying occurrence of behaviour
- 5What are the 2 types of reinforcement in operand conditioning?Positive: favorable stimulus after behaviour
Negative: removal of aversive stimulus after behaviour - 6What is the main principle of Hebbian learning?CELLS THAT FIRE TOGETHER, WIRE TOGETHER
- 7What is the difference between cells firing together and separately?Cells that fire together STRENGTHEN synapses,
Cells that fire separately LOSE synapses - 8What is LTP and LTD?Long Term Potentiation;
Long Term Depression - 9Features of LTPIncrease in EPSP, long term synapse STRENGTHENING
- 10Features of LTDIncrease in IPSP, long term synapse WEAKENING
- 11What are the properties of Associative LTP?COOPERATIVITY (works when >1 input is simultaneously active),
SPECIFICITY (only stimulated synapse shows potentialisation),
ASSOCIATIVITY (weaker inputs are potentiated when co-occuring with stronger inputs) - 12What is the NMDA receptor responsible for?Central to producing LTP, but not maintaining it
- 13What are the steps to producing LTP?1. NMDA receptor opens
2. Glutamate binds
3. Post-synaptic cell depolarizes
4. Mg2+ is ejected out of the channel
5. Ca2+ enters and facilitates LTP - 14What are the 3 types of learning in machines?Supervised Learning
Assisted Learning
Unsupervised Learning - 15What are Artificial Neural Networks (ANN)?Very crude approximations of biological neural networks
- 16What do ANNs consist of and what do those things have?They consist of nodes and connections between these nodes;
nodes have input/output and can perform simple computations like Action Potential.
The connections have a weight. - 17What are their types of general structure?Fully connected, feed forward (one connected to one other, but not the other way), Reccurent network (some back connections)
- 18What does the perceptron do?Evaluates different inputs, then given the weights it gives the best configuration of all the inputs; this is how it learns and comes closer and closer to the best configuration.
The configuration tells us how much we should use of each input optimally. - 19What is back-prop?Training the ANN to adjust to a certain gradient
- 20What are the features of back-prop?It can be generalized to any number of layers,
it's only feed forward - 21What is the back-prop in rules?Weight - Change = some small constant x error
- 22What are the disadvantages of back-prop?Learning can take a long time, new programs will overwrite old ones, can't be overtrained
Information
- Description
- Flashcards on Cognitive Psychology: Learning, created by krisbraz on 12/12/2014.
- Created by
- krisbraz
- Views
- 0
- Rating