PSYCH 85 FINAL Flashcards

1
Q

LEFT BRAIN

A
Mostly Language
-Grammar
-Naming
-Repeating
-Understanding
Verbal Memory
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

RIGHT BRAIN

A

Attention
Spatial Processing
Faces
Nonverbal memory

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

How does handedness relate to dominant hemispheric processing of language?

A

L Hemisphere processes language. A person’s handedness is opposite from their dominant hemisphere. Left handers had language processes more dispersed through their brain

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

How does hemispheric damage affect the emotional reaction to brain damage?

A

Brain damage to the left hemisphere is more likely to be catastrophic;
Result in despair, hopelessness, or anger

Brain damage to the right hemisphere is more likely to result in an indifferent reaction; Euphoric reaction–Minimization of symptoms, placidity or elation

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Which side of the brain causes catastrophic response?

A

LEFT BRAIN. Results in despair, hopelessness, or anger.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

-Minimization of symptoms, placidity or elation

How do we assess brain functions of a hemisphere in normal people?

A

WADA test

  • Inject anesthetic (sodium amytol) into the right or left carotid artery
  • Puts one hemisphere to sleep so we can see what functions there
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What is hemispatial neglect and how likely is it given damage to each hemisphere?

A

Hemispatial neglect: A failure to report, respond, or orient to stimuli presented contralateral to the side of a brain lesion in the absence of elementary motor or sensory deficits

Most often occurs with right hemisphere damage, loss of left visual field

  • See everything on right
  • Not just a sensory problem but rather a problem of consciousness
  • 86% with Right hemisphere damage
  • 7% with left hemisphere damage
  • 7% with bilateral damage
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What is anosognosia and anosodiaphoria?

A

Anosognosia: Lack of awareness or denial of any problem

  • Left arm paralyzed, right are is fine, doctor says touch nose with left arm, uses right arm to touch nose, says that they are touching nose with left arm, says they can visually see they are touching their nose
  • Perceiving something wrong

Anosodiaphoria: Awareness of deficit but without appropriate concern

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

appropriate concern

What is alien hand syndrome? How does the mirror box treat phantom pain? How does the cortex reorganize after losing a body part?

A
  • Inability to control one hand
  • Hand can perform complex behaviors (Like buttoning a shirt)

Mirror box:
Normal arm in one slot
Mirror image is superimposed where phantom arm would be
Move real arm until matches position of phantom limb
Close their eyes and make symmetric movements then open eyes
4/5 subjects with involuntary clenching spasms found relief
Temperature did not transfer (control for confabulation)

Cortex can still feel lost limbs like they are “phantom”
-Eventually cortex reorganizes to other parts of body

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What is declarative memory and what are the brain structures that process declarative memory? What is retrograde and anterograde amnesia? How does memory consolidation work? What kind of memory deficits does consolidation produce?

A

Declarative/Explicit memory associated with Hippocampal formation in medial temporal lobe
Declarative memory is the memory of facts and events

Retrograde Amnesia: Can’t remember things in past
-Temporally graded: Forget things that happened just before accident occurred
-Harder you hit your head the more you forget
Why is it temporally graded?
-Hippocampus formation holds info for a while and “teaches” info to the rest of cortex
-Teaching is part of consolidation
-After new info has been “taught” then is is stored in rest of cortex
-Info that is not taught yet, is lost

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is apperceptive and associative agnosia? Where are faces processed in the brain? What is a face-processing deficit called?

A

Apperceptive Agnosia: Cannot assemble parts into a meaningful whole
Associative Agnosia: can perceive but can’t label

Faces detected in fusiform area
Inability to recognize faces = prosopagnosia

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What is neural synchrony? Describe different types of neural synchrony.

A

How does shape and color get “bound” together
Simple: Firing together at the same time
Complex: Firing in similar patterns
-May be responsible for much more than binding of perception
-Memory?
-Consciousness?

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

How can temporal binding be used to represent object shape?

A

Temporal binding: Combines what we’re processing with where we’re processing it

Imagine you’re looking at a suitcase, which has a rectangle/cube unit (the body of the suitcase) below a tube unit (the handle). Maybe the neurons for the cube units fire at the same time as the “below neuron” and the neurons coding for the arc units fire at the same time as “above units” to bind the parts with their relative location

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What are some similarities and differences between the brain and standard computers (like the ones running a typical pc)?

A

Similarities:
Both store and use info
Both have working memory (Computer = RAM ~8-12 GB, much smaller in humans)
Both have long-term memory (Computer = Hard-disk, CD)
Both have control structures
-Computer = CPU
-Man = Attention

Is there a central part of the brain that directs all other parts of the brain? What controls the brain?
Not completely centralized, much more distributed than a CPU

Differences:
Brains:
-Distributed parallel power, like a billion little computers (relatively slow, but if one brain cell dies, rest of brain is unaffected)
-Fault tolerant (tough to break)
-Very good at learning: Making associations, learning new patterns, probably biggest strength
-Coded in fuzzy analog format (Analog: Means you can take any value, means you can code for any number, not just 1s and 0s)
-Distributed Representation (Distributed: Info is stored between a lot of different neurons)

Computers:

  • Usually 1 serial processor (super fast, if processor dies, computer dies)
  • Very sensitive to damage (fragile)
  • Not naturally suited to learn (super FAST, not meant to learn)
  • Coded in 1s and 0s (binary: need to be super careful while creating program, otherwise program won’t work)
  • Local representation (Local: computer has all info stored in one location)

Can computers learn? Sure

  • Take browser history to suggest ads
  • Only possible because a ton of programming makes it happen
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What are local and distributed representations?

A

Local Representations: code a concept with one node

Distributed representations code information by a pattern of activations across a set of units

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is the basic components of a neural network?

A

Network has nodes (kind of like neurons)-Node: Unit that takes on certain activations
Network has links between nodes (like connections between neurons) -Passes activation from 1 node to many other nodes
Nodes have levels of activation–Activation rules: How we combine activations/how activations spreadd
Links between nodes have connection strengths (weights)-Can have strong excitatory (positive), near neutral strength (0), and strong inhibitory (negative)

Summing inputs:
Multiply strength at A by strength at C and add the product of the strength at A* the strength at B to ged the total activation at A
(see slide 9 of lecture 13)

17
Q

What is the basic structure of a perceptron? What are it’s limitations? How can those limitations be overcome?

A

?
First artificial neuronal model
Have 2 inputs that are binary (1 or 0)
Think of input A as true or false
Think of B as another statement T/F
True = 1, False = 0
1. Entire statement is true is A AND B are true (If A = True, B= False, system - false, if A = false, B = true, system = false, if A = false, B = false, system = false, if A = true, B = true, system = True) NEURONAL NETWORK SOLVED IT (Just need to add up inputs to get > 0)
2. A OR B (If A = true, B = false, system = true, if A = false, B = true, system = true, if A = true, B = true, system = true, if A = false, B = false, system = false)
NEURONAL NETWORK SOLVED IT (Just need to add up the inputs to get > 0)
3. A XOR B (Exclusive or)
A can be true
B can be true
BUT NOT BOTH
Neuronal network couldn’t solve this problem because you can’t just add up the inputs

Neuronal network can only solve problems that are linearly separable (can draw a line between what is true and what is false)
Perceptron cannot do:
-Exclusive OR
-Even/odd discrimination (even or odd number)
-Inside/outside discrimination (point inside or outside shape)
-Open/Closed discrimination (open or closed shaped)

Can be fixed by having multiple layer perceptrons
2 layers, easy to make network learn, clear pattern between input and output layer, very limited in what network can learn

3rd hidden layer can solve many more problems, comes at a cost, makes it much harder for network to learn

18
Q

What is supervised learning? What problem is there in applying supervised learning algorithms to the human brain?

A
  1. Feed network inputs
  2. See what activations become
  3. Compare this to what it should be (desired output)
  4. Change the weights between nodes according to how much error they contribute

Problem in human brain: Where does this teacher come from?

19
Q

What are some desirable characteristics of neural networks?

A
  1. Distributed Representation: Ideas, Thoughts, concepts memories are all represented in the brain as patterns of activation across a large number of neurons. As a result, there is a lot of redundancy in neural representation
  2. Graceful Degradation: Performance of the system decreases gradually as the system is damaged (As neurons get knocked out, entire system doesn’t get destroyed
  3. Learning: Delta Rule/Backpropagation
  4. Generalization: Because of how the network learns and its distributed representation, it can respond to inputs it was never officially trained on, generalizing based on similarity to things it was trained on
  5. Distributed processing: Not only representation, but processing is distributed too, so there is no central controlling function, or CPU, in the brain. It is more cooperative. (Problem with distributed processing for computers is software)
20
Q

What are some problems with creating artificial neural networks?

A
  1. Stability-Plasticity dilemma: Learn new info while retaining old info (want to make sure that new stuff doesn’t affect old stuff
  2. Catastrophic Interference: System falls apart when new info is learned (learn something new, everything shuts down)
  3. Supervised networks (Where is the teacher?)
  4. Biological plausibility of teacher

Supervisor: Knows what the output should be, allows us to check work, see where things went wrong
In human world, don’t always have supervisor

21
Q

What is the concept of centrality in the discussion of networks? What are some examples of centrality?

A

Computers-CPU
Armies-Generals
Politics-Presidents and dictators
Human Mind?-Can’t pinpoint area of brain that controls everything else

22
Q

Give an example of a hierarchical network in the human brain.

A

Simple cells: Oriented bars
Complex cells: Oriented Bars moving
Hypercomplex cells: Right angle of vertical and horizontal lines moving in the same direction

The further you get into brain, the more cells you get that can detect complex info

23
Q

What is a small-world network? What is a random network and an ordered network?

A
  • 6 degrees of Separation
  • 4 degrees of Kevin Bacon (Takes only 4 steps to get to any American actor to any other American actor)
  • Electrical Power Grid
  • Railroad
  • Nervous systems of many animal

Random networks: Local AND Global links, can connect to any node
Ordered networks: Only local links, only connects neighboring nodes
Ordered networks are more protected but harder to get to different parts of the network

24
Q

What is an egalitarian and an aristocratic network?

A

Egalitarian network: Links are evenly distributed

Aristocratic: Some hublinks are especially important (like google on the internet)

25
Q

What is percolation? What psychological disorder might exhibit percolation? How might percolation be beneficial for certain types of thinking?

A

Percolation: Spread of disease in a network of people
Schizophrenia exhibits percolation: Disorganized thinking, spread of activation similar to a percolating cluster

Percolation could be beneficial in Divergent thinking: Helps you generate a creative output

26
Q

What are structural and functional kinds?

A

Structural Kinds:
-What is is made of?
-Medium dependent
For human minds, the material that it is made of plays a role in what it can process
-ie. Can’t process UV light because we can’t see

Functional kinds:

  • What does it do?
  • ie. Mouse trap
  • something that traps mice
  • What is is made of? ANYTHING THAT TRAPS MICE
  • Nuclear bomb could be extreme case of mouse trap
  • Defined by what it does, not what it is
27
Q

What is the Turing test? How does it relate to functional and structural kinds? What are some criticisms of it and what are some responses to those criticisms?

A

Turing test asks: Can someone be another person?
Interaction: Can only see the words typed out
Can ask as many questions as your want
Can you distinguish between a human and artificial intelligence
Not really a test of intelligence, but rather a test of human intelligence
Computers used to be at ~.01% of what we consider intelligence
As time goes on and technologies advance, that number becomes higher and higher
If it reaches 100% computers would be smarter than us at everything

People have been trying to create devices that are smart enough to pass the test
Puts emphasis on the judge
The smarter/better the judge is, the harder the test gets
If the judge does not know that much, than it is really easy to pass the test, if the judge is really smart, it becomes difficult to pass the test

Criticisms:

  1. The judge might be an idiot
  2. All Possible conversations program–Talking so much that they might answer every possible question
    - If you program all the types of questions and all the types of answers
    - Practically possible and impossible
  3. The too-smart argument (computer can respond too smart to be a human ie. too complex math)
  4. Searle’s Chinese room Argument (Syntax vs. Semantics: Correct output vs. understanding what you’re saying)
28
Q

What is Searle’s Chinese room? How does it relate to the Turing test?

A

Man in room asked a question in Chinese. Doesn’t understand Chinese but uses rule-book to write correct answer.
Man in room lacks intentionality
The rule book captures intentionality, had to be created by a human mind

Just mapping input to output for speech. Lack of semantics, intentionality
-Artificial intelligence can correctly answer questions of Turing test but doesn’t know meaning behind answers

-Responses to Searle: Systems Argument
He could memorize the rule book and internalize the entire system
-Responses to Searle: Robot Argument
If interacted w/ world, would know Chinese but even if a robot had legs, still just manipulating symbols
-Responses to Searle: Back to Future Argument
Rule book had to come from a mind so you are actually talking to that mind back in time
-Responses to Searle: Probability Argument
Generating a system is impossible

29
Q

What is a Turing Machine? What can it do?

A

Develop a machine that can compute ANYTHING that it is logically possible to calculate
Foundation for modern computer. Strip of tape with given set of rules.
State –> input –> instruction (write, move, move, stop)
Simple: tic-tac-toe
“a mathematical model of a hypothetical computing machine that can use a predefined set of rules to determine a result from a set of input variables”

30
Q

What are some examples of recent developments in artificial brains and/or intelligence?

A

IBM 2014 Chip
Parallel computing (4096 computers)
46 billion operations per second
Equivalent to 1,000,000 neurons (Bee brain)

Mouse Brain memory implant
Tagged memory cells by observation Used optogenetics to implant associations
Brain-like tissue comprised of collagen and silk proteins

31
Q

Give some examples of how biased sampling can impair judgements.

A

Early Dating
“Love” chemicals biased sample
Your behavior is biased sample of you
Your partner’s behavior is biased sample

Playing with other people's kids--So fun, don't have to deal with bad, might make you want to have a kid
Fighting with someone you love--Angry, immediately think of all the bad things, might make you want to break up
Job interviews--Presenting your best self, so is company, might convince you you would really wanna work there even if its not really for you
Inspirational speeches (Steve Jobs) Politicians--Reach for the stars, might be inspired to go out on limb, take risks
People talking about politics--Might become overly defensive, cause overreaction
Evaluating your own behavior