Neural Networks and Cognitive Control

Question 1

3 goals of science

Answer 1

1. description: what are we observing
2. prediction: what will we observe next
3. Explanation: why is that what we observe

Question 2

what was the early explanation of planetary motion

Answer 2

ancient greece: helios driving his chariot- when ever he was in the mood
- not a goo d prediction

Question 3

3 models of planetary motion after helios

Answer 3

• ptolemaic geocentric model
• copernican heliocentric model
• keplers laws of planetary motion

Question 4

keplers law of plantetary motion accounted for all: (3)

Answer 4

1. description: observed data
2. Prediction: extremely accurate predictions
3. explanation: ‘simple’ elegant framework
   - -> the things we want in our science

Question 5

why do we have quantitative models? (7)

Answer 5

1. data requires a model to be understood and explained
2. verbal theorizing alone cannot substitute for quantitative analysis
3. there are always several alternative models that vie for explanation of data, and we must compare those alternatives
4. model comparisons rests on both quantitative evaluation and intellectual and scholary judgement
5. even seemingly intuitive verbal theories can turn out to be incoherent or ill specified
6. only instantiation in a quantitative model ensures that all assumptions of a theory have been identified and test

Question 6

what is the fundemental trade off with models

Answer 6

• more detailed = more accurate but hard to understand
• have to make it specific to what you are looking for
• –> simplicity & elegance versus complexity & accuracy

Question 7

what is the goal of modelling

Answer 7

maximize explanatory power while minimizing complexity

Question 8

what types of cognitive models are there? (4)

Answer 8

• mathematical models (fitts law)
• symbolic models
• dynamic systems models
• hybrid models

Question 9

ways you can wire neurons is called

Answer 9

topology

Question 10

4 types of topology models

Answer 10

1. feedforward
2. simple recurrent (elman)
3. self organizing map (kohonen)
4. fully recurrent

Question 11

adjusts weights based on difference bewtween actual output and correct output (shown an object and you say what it is an dif you are right you are told and if wrong you are told, you can adjust)

Answer 11

supervised learning

Question 12

adjust weights based on correlations in input “neurons that fire together wire together” (keep exposing it to lots of input and it learns what things go together)

Answer 12

unsupervised learning

Question 13

adjusts weights based on difference between actual reward and expected reward

Answer 13

reinforcement learning

Question 14

a feedforward model is what type of learning? how does it learn weights?

Answer 14

supervised

- back propagation

Question 15

a single layer hopfield network is what type of learning

Answer 15

unsupervised learning (unsupervised recurrent networks)

Question 16

kohenen map is what type

Answer 16

unsupervised recurrent

Question 17

elman- words and letters is what type of map

Answer 17

supervised recurrent network

Question 18

what is the RL problem

Answer 18

learn what action to take to maximize total reward given the current state of the environment

Question 19

what is cognitive control

Answer 19

its about guiding and coordinating all of these processes in service of our plans and goals (operating system of our mind)

Question 20

cognitive control is sometimes referred to as?

Answer 20

executive control, central executive or self control

Question 21

what is the homunculus? what is the goal of cognitive control?

Answer 21

• the little self in their that control us
• banish the homunculus: step appealing to the ill defined and circular self, central executive or consciousness to explain goal orientated behaviour
• goal: a mechanistic account of cognitive control in neural, psychological and computational terms

Question 22

what are the components of cognitive control?

Answer 22

1. shifting between tasks or mental states
2. updating and maintaining working memory representations
3. inhibition of dominant or prepotent responses
4. monitoring and adjusting responses

Question 23

what lateral prefrontal areas are involved in cog control?

Answer 23

dorsolateral, ventrolateral

Question 24

what ventromedial prefrontal areas are involved in cog control?

Answer 24

orbital frontal

Question 25

what medial prefrontal areas are involved in cog control?

Answer 25

dorsomedial, ACC

Question 26

schiz effects about __% of canadians
and is _____ times more frequent in females than males
and peak age of onset for females is ___ and ___ for males

Answer 26

• 1%
• 1.4
• 27, 25

Question 27

positive symptoms of schiz

Answer 27

hallucinations, delusions, thought disorders, movement disorders

Question 28

negative symptoms of schiz

Answer 28

disruption of normal emotions and behaviours

- flat affect, reduced feelings of pleasure, difficulty beginning and sustaining activities, reduced speaking

Question 29

cognitive defecits in these areas in schiz

Answer 29

• attention
• working mem
• episodic mem
• executive function