Midterm Flashcards
(117 cards)
1
Q
Scholasticism
A
The philosophical systems and speculative tendencies of various medieval Christian thinkers, who, working against a background of fixed religious dogma, sought to solve philosophical problems with intellect, realism and nominalism. Merged classical thinking and Christian doctrine. Science still not applied to the mind.
2
Q
Roger Bacon (1620)
A
Father of scientific method
Not focused in religion, goal was to improve life/humanity
3
Q
Materialism
A
Belief that one type of “stuff” makes up the universe. Not liked by the Catholic Church, who thought it was anti-religious
4
Q
Dualism
A
- Separation of the mind and body
- Offered some explanation of how cognition was not materialistic
5
Q
Bell-Magendie Law (1811)
A
- Charles Bell
- Spine is the cause of all sensation and movement
- Per church: physical explanation for souls
6
Q
Law of specific nerve energies
A
- Johannes Müller
- The mind has access to things that are only in the nerves (not real world)
- The contents of the mind have no qualities in common with environmental objects but serve only as arbitrary signs or markers of those objects.
7
Q
Applying Materialism to Mental Life: 2 Interpretations
A
- Materialism + Evolution → the adaption of the brain over the lifetime of the individuals (Herbert Spencer, Herman Ebbinghaus)
- Materialism + Methods → Consciousness as a subject matter (William Wundt)
8
Q
3 schools of Psychology in late 19th century
A
Structuralists, Wurtzberg School, Act psychology
9
Q
Structuralists
A
- Wundt,
- Method: introspection
- Focused on breaking down mental processes into the most basic components
10
Q
Wurtzberg School
A
- Kulpe
- Methods: mixed.
- Hypothesizes existence of special states of consciousness—“thoughts”—which cannot be reduced to the sensory content.
11
Q
Act psychology
A
- Brentatno
- More philosophical than empirical
- The mind is a symbol system
12
Q
William Wundt
A
- Recognized as the founder of psychology
- Voluntarism (Wundt’s experimental psychology
- Volkerpsychologie (Wundt’s non-experimental psychology)
13
Q
Voluntarism
A
- Wundt’s experimental psychology
- Meant to indicate voluntary, active, and willful nature of the mind
- Key was Apperception
14
Q
Apperception
A
- Active intentional process involving will
- Mechanism of creative synthesis by which psychical elements and compounds are synthesized into experience
- Key to Wundt’s Voluntarism theory
15
Q
Volkerpsychologie
A
- Wundt’s non experimental psychology (cultural psychology)
- Precursor to social psychology cultural psychology, and personality
16
Q
Wundt’s goal of psychology
A
To analyze experience in terms of component elements and compounds
17
Q
Two types of experience
A
- Mediate experience: domain of natural science
- Immediate experience: domain of psychology
18
Q
Experimental self-observation
A
- Wundt’s self described experimental methd
- Observer presented with stimulus condition and instructed to be in a state of readiness
19
Q
Tridimensional theory of feeling
A
Wundt’s theory that all emotions take place on three separate continua: pleasant-unpleasant, tension-relaxation, excitement-depression
20
Q
Ebbinghaus
A
- Forgetting curve
- Wanted to study associations as they were being formed
- Nonsense syllables method
21
Q
Systematic experimental introspection
A
- Observers would experience what ever stimuli or events they were supposed to experience and then provide comprehensive account of mental processes
- Pioneered by Kulpe
22
Q
Imageless Thought
A
- Pioneered by Kulpe
- Observers reported forming images of weights when lifted them but reported there was no sensory or imaginal content present when they made judgments
23
Q
Gestalt Psychology
A
- Koffka, Wertheimer, and Kholer
- Sought to explain perceptions in terms of gestalts rather than by analyzing their constituents.
24
Q
Behaviorism
A
- Goal to predict behavior or show how classical and operational conditions can account for behavior
- Methods cannot be subjective, introspection as a method in invalid
- Pioneered by Watson
25
Neobehavioraism
- These psychologists were interested in theory, focusing their research on learning and motivation
- Neobehaviorism (muscle twitch psych)-reflex is a functional relation
- Hull, Skinner, Tolman
26
Introspection Method Goal
To find the basic units of mental life underlying consciousness
27
Cognitive Method Goal
To figure out which representations and which processes the mind uses
28
What was Behaviorism unable to account for?
Ethology and Language
29
Cognition Movement
- The goal is to account for behavior, but the focus is on mental life as unconscious
- Method: Explanation = processes that manipulate internal representations
30
Principles of Cognition
-Symbols replace S-R/conditioning as the unit of thought
- Mental life is unconscious
- Intervening variables could exist
- Existence of hypothetical constructs (e.g. strategy, mental images)
31
What was Cognition unable to account for?
How can we avoid “tricking” ourselves if we are allowed to make up what a mental representation is?
32
Four positions on cognition
1. Study computation without the brain
2. Study computation but leave out cognition
3. Start with mapping theory
4. Brain is involved in construct development
33
Informational description
- Related to 4th position on cognition
- Mental events can be described in terms of “input” “operation” and “output”
34
Recursive decomposition
- Related to 4th position on cognition
- You can decompose each even into a more basic event, but at some point you hit a primitive that is not further reducible at a functional level
35
How did cognitive psychologists respond to the issue of “made up” constructs
- Neural constraints
-Formal models
- Evolution
- Behavioral measures
36
Two types of long-term memory
Type - enduring representation
Token - duplicate of the representation you create to manipulate in your mind
37
Information Theory
- Quantitative approach to psychology
- Was meant to be a way of thinking of human information processing that was going to constrain theory
- Logarithmic relationship between information and reaction time
- Way of measuring information irrespective of content
- Valuable as one way of describing the environment
- Mapping theory
38
Donders Subtraction Technique
Perception and motor time = time required for simple task
Discrimination time = time for discrimination task minus simple task
Choice time = time for choice task minus discrimination time.
39
Miller's Law
The number of objects an average human can hold in short term memory is 7 + 2
40
Key Features of Cognitive Strategy
- Develop alternate methods of processing
- e.g. search (serial: exhaustive, serial: self-terminating, parallel)
- Derive signature predictions of each models
- Obtain data that allow comparison
41
Mental Chronometry
- Discrete stimulus = Task C - Task A
- Response Selection = Task B - Task C
42
Sternberg's Additive Factors
Use positive sets to as base stimuli
Subject responds if a number is in the set
Step 1: Specify a model
Step 2: Identify the stages with different IV and DVs
43
Independence problem in additive factors
Method assumes that each step of the model is independent
44
Signal Detection Theory
Testing how far a patient falls from the line between hit rate and false alarm rate
45
Types of Cognitive Methods
- EEG
- Single Cell Recording
- fMRI
- Testing patients with lesions to the brain
46
Single Cell Recording
- Records the activity of a single neuron in an animal
- Method: place a wire through a hole made in skull
- Benefits: precise temporal information and location of information (after repeated testing)
- Issues: highly invasive-not done on humans
47
EEG
- Pros
- Noninvasive
- Can be used in little kids
- Temporal resolution is great
- Can look at interaction of processes
- Inexpensive
- Cons
- Most confident about cortical results
- You need a lot of trials
- Real possibility of carryover effects
- Using the same stimulus set
- Not very interpretable to people outside of the field
48
fMRI
- Pros
- Spatial and temporal resolution are good
- You can do an anatomical span
- Cons
- Cost
- Things messing up the signal
- So much data that you can do shady data processes or fool yourself
49
Four Methods of Using Neurological Data
1. Single Cell
2. Use before theory development to guess what is happening
3. Look at brain to confirm theory
4. Use brain data to add validity to a psych theory
50
Evolutionary Psychology
Initially intended to analyze cognitive abilities in tandem with likely selective pressures and reverse engineer development
51
Task analysis
- David Marr
- Cash register analogy- take something you can’t see the insides of and look at output-deduce what function is from behavior
- Evolutionary psych aims to look at output and deduce how it is key to survival
52
Franz Brenatno
- Founded Act psychology, molar psychology that called for a larger unit of analysis in looking at consciousness
- Foreshadowed American functional psychology and Gestalt psychology
- Originated idea of "Experimentium Crucis"
53
Experimentium Crucis
- Proposed by Brentano
- Science best served by a few grand experiments testing big questions
54
Oswald Kulpe
- Structuralist, Wertzberg School
- Came up with systematic experimental introspection, imageless thoughts, and mental sets
55
Survival Memory Effect
- Implant bias judgment- word test then memory test
- 1 group makes word judgements based in likeability, the other makes judgements based on use in wilderness survival (on a grassland), one makes judgment about survival in a new city
- Increased memory occurs in grassland condition
- Nairne
56
Criticisms of Evolutionary Psych
- Post-hoc explanation of findings - finding causal explanations for phenomenon
- Swiss Army knife phenomena not good because assumes one does not impact others
- People who apply evolution to psych are not biologists
- Once you get to technology and culture, things get much more complicated
57
Driving force behind mathematical psychology
William Estes
58
Purpose of computational modeling of psych
- Reduce and order data
- Qualitative data → quantitative data (Scaling theory)
- Quantify qualitative models
- From the world we abstract a model, from the model we derive a prediction, from the prediction we interpret data.
59
Model fitting
- Metric that includes the quality of the fit and the ratio of free paramaters to data accounted for
- Not additive
- What you need to do is model comparison
60
Diffusion model
- Model of choice for reaction time
- Stimulus encoding process- each count of time (millisecond) is information accumulating
61
Why is it better to compare mathematical models rather than just make them?
- Comparing models forces specificity
- Makes models more concrete and makes sure you know what is going
62
Advantages of computational models
- If you are modeling something complex, very difficult to keep in your head
- Computational models are explicit
- Computational models force you to be specific
63
Criticisms of computational models
- Frequently not clear what central features of a model are just by looking at the model
- Once models get complicated, very difficult to assign meaning definitively
- Number of free parameters is an unresolved issue
64
Intentionally
How representations come to bear meaning
65
Grounded representations
Representations that are perceptual or motor
66
Who investigated the functional role of representations?
Arthur Markman
67
Four types of representation
1. Spatial Representation
2. Feature Model
3. Semantic Network
4. Structured Representation
68
Pros of Spatial Representation
- One dimension of space for what we are trying to represent
- Uses continuous units
- Relationally obvious
- Can change things or add mechanisms to a spacial model
- Can stretch or compress
- Simplicity of model is best aspect
69
Cons of Spatial Representation
- No direct access to to features that made judgment
- Model is useful when speed is prioritized over detail
- All dimensions play and equal role unless specified
70
Pros of Feature Representation
- Uses discrete unites
- Features stay determined
- Example: Cold/Flu with symptoms represented as (+) or (-) for presence or absence.
71
Cons of Feature Representation
- Not clear about what factors should be added
- Compromise between spatial and semantic network
72
Pros of Semantic Network
- Very good at relationships among concepts and associations
- Example: Network analysis
- Good at relations where they are explicit
73
Cons of Semantic Network
- People also don't describe objects with hierarchy in mind
- Power (too powerful)
- 1 node per concept does not allow for wrong information or lack of info (must know everything about everything in concept)
-Does not allow for slow separation of representativeness (assimilation-accommodation)
74
Semantic Network: Marking Passing
- Tried to deal with choosing features
- Only specify as needed
- At base-only relevant to one representation
- Efficiency os also considered
- Made when we were concerned about brain space
75
Semantic Network: Spreading Activation
- Weights are added to connections
- Labels are lost
- Activation is graded in representations
- Example: Network analysis
76
Pros of Interactive application model
Deals with multiple constraints simultaneously
77
Pros of cascading model
- As some parts become active, they reach down and start inhibiting other representation
- Uses top down processing to constrain likelihood of lower process of representation
78
Structured representation
- Rich information about how how parts of representation relate and work as a part of the whole representation
- Takes an argument and varies kinds of arguments or relationships
- Come from AI and computer intelligence
- Restricts the scope of a represational element
- Three type examples: Schemas, Frames, Scripts
79
Schemas
A cognitive framework or concept that helps organize and interpret information.
80
Script
A form of memory structure that evolve over multiple exposure to the same set of stimuli and/or repeated enactment of a particular behavior.
81
Parallel Processing Models
A set of system in which memories are stored and retrieved in a system consisting of a large number of simple computational elements, all working at the same time and all contributing to the outcome.
82
Distributed Models
- In distributed models you have to look at contributing units to understand a pattern
- Uses “neuron like” elements
- Good at pattern completion (e.g. memory is a pattern completion tool)
- Assign default for missing features
- Output gets corrective feedback and slowly changes
83
Weaknesses of Distributive Models
- Catastrophic interference
- Not good at sequential modeling
- Cannot explain rule based behavior
- Neuron like elements oversell the connection to biology/neruo
- Computational power is debated
- Hard to understand
84
4 Common Types of Categorization
1. Classical/rule based view
2. Prototype view
3. Exemplar view
4. Category/boundary view
85
Classical/rule based view
- Category is an object
- Concept is a mental representation of category/necessary and sufficient features
- Example: Trunk, big ears, tusks = Elephant
86
Prototype View
- Categorization is the central tendency of all the different members of the category
- Concept = a representation that has all the features that are typical of the concept
- Computationally efficient but less practical
- “Typical” object is represented and compared with other prototypes
87
Exemplar View
- Structure/process trade off
- New stimuli were saved amongst all seen examples
- Can be used to make a prototype
- See at encoding → store each exemplar → derive prototype if needed
88
Category/Boundary view
- The distance from a boundary explains the the similarity/typicality of a result
- The distance from a boundary explains the the similarity/typicality of a result
- Compliment to prototype view but looks at periphery rather than the category itself
89
Issues with Similarity Model
- Dependent of self selected features (we identify what seems important)
- Sometimes certain features have outsized importance
- Rules matter, and exceptions exist
90
Rule X Model
- Derived by Nosofsky
- People are lazy and just do what is easiest
- They are likely to made a general rule and memorize exceptions
- Perceptual categorization is really important for experts
91
Role based categorization
- Also called categorization expansion
- Categorization is based on relationship with other categories
- Focused placed on extrinsic rather than intrinsic factors
92
Frames
- Schema that helps organize a situation
- Helps attend to relevant features
93
Propositions
- Ways of storing complex information in schemas
- Can predict connections between words
94
4 main cues for connecting propositions
- Similarity (about the same thing)
- Connectivity (key word connecting the sentences)
- Causality (one thing causes another)
- Setting (things happening in same place)
95
How does the human brain tend to identify objects?
Shape
96
Inverse projection problem
Three dimensional space can project an infinite number of 2 dimensional images on the retina
97
Viewpoint independence
We are pretty good at identifying objects from different viewpoints
98
4 Types of Model Features
1. Primitive elements
2. Reference frame
3. Relations of primitive elements
4. Parts vs. whole
99
How do we know what things are from different perspectives?
4 types of model feature theories
100
Primitive Elements
- Vertices/lines or geometric solids
- We are better identifying things with vertices intact than lines intact
101
Reference Frame
- Viewer centered
- Object centered
- Hybrid
102
Relations of primitive elements
- Coordinates in space
- Relative relations like above/below
- Sometimes called categorical
103
Parts vs. whole
- Analytic vs. holistic
- Analytic: different parts of elephant together = elephant
- Holistic: face = face (parts don't look like anything on their own - knowing this is a face depends on spatial relationship of objects).
104
View Based Model
- Vertices
- Viewer-centered
- Coordinates in space
- Holistic
105
3D Matching Models
- Lines, gemetrics solids
- Hybrids
- Coordinates in space
- Analytic
- Matching what is out in world into back projection
106
Structural Description Models
- Geometric solids
- Object centered, viewer centered, hybrid
- Relative relations like above/below
- Analytic
107
Biederman's Model
- Addresses viewpoint independence by introducing geons
- Too effective
- Can ID shape from obstructed view but too well
108
Why is it difficult to compare models of object recognition?
Any object can be recognized if you have enough 2D representations of it
109
Tarr's viewer centered experiments
- Used geons to show that the model expected better reconciling of viewpoint independence than was true
- Found that people could not really ID objects in space if not trained on them
110
Geons
Shapes of different rotations that could form different common objects
111
Categorical verus coordinate change in object recognition
Studies suggest we are more sensitive to categorical change in objects
112
Two viewpoints in visual object recognition
- Psychological reality and the brain
- Brain as a model system
113
Problems with DCNN
- Need more varied training that is more similar to how humans are trained visually
- Don't compare to human data but rather how we see neurophysiology respond
- Do not account for cortical Specialization
114
Deep convolution neural networks (DCNN)
- Receive images as an input and use them to train a classifier
- The network employs a special mathematical operation called a “convolution” instead of matrix multiplication
- The architecture of a convolutional network typically consists of four types of layers: convolution, pooling, activation, and fully connected.
115
Flow Diagram of Attention
- Broadbent
- Post WWII examining control tower operators attention (1950’s)
116
Is Taxonomy effective in the study of attention?
No
117
