Please enable JavaScript.
Coggle requires JavaScript to display documents.
Chapter 1: Main Approaches to Studying the Mind (Sessions September 7…
Chapter 1: Main Approaches to Studying the Mind
(Sessions September 7 and September 12)
Cognitive Psychology
:star:
Controlled behavioral experiments with healthy participants in laboratory conditions
Main Assumptions
Information-processing approach:
external stimuli cause internal processes to occur
Top-down
versus
Bottom-up
processing
Metaphor: the mind as a computer.
Serial
versus
Parallel
processing
Encoding, memory, responding;
Discrete stages
vs.
cascading
human information processing is active,
not just passively taking up information
and forming associations
mental processes can be inferred from
human behavior in carefully controlled
(experimental) settings
Reaction Times
Cognitive processes take time
time from a stimulus to a response in a given experimental setting
estimated duration of component processes by comparing reaction times
Donder's Subtractive Method
S = stimulus
R = response
RT = reaction time (in msec)
A-Task: A simple reaction time task - would require perception and motor stages - time to receive and then execute the stimulus.
NSCT = Nervous System Conduction Time
(baseline reaction time)
C-Task: A discrimination reaction time task requires the above plus a discrimination stage.
B-Task: A choice reaction time task - requires all of the above - time to receive and execute the stimulus, and discriminate plus a choice stage.
Metaphors and analogies
Theories and models. A theory is a set of related statements to explain a phenomenon. (Verbal theories, mathematical explanations, computer programs)
Falsification: results of an experimental/empirical study can show that a theory is false,
but never that it is "true"
Experiments
manipulate stimuli
,
observe responses
,
infer internal preocesses
:red_cross: Limitations:
Task impurity problem
Lack of ecological validity
Vague theories - hard to make predictions from them
Paradigm specificity: findings don't generalize across tasks
Strengths:
first systematic approach to understanding cognition
produced numerous theories and tasks adopted by other approaches
Experimental Designs
IV, DV, Neutral baseline
repeated observations within each condition and participant, averaging RTs across repeated observations
Factors
Interactions
Cognitive Neuropsychology
:star:
Patterns of behavior shown by patients with brain damage
Lesions
: structural alterations of the brain caused by injury or disease
Its principal aim is not learning about the brain, but about the mind
Main Assumptions
Functional Modularity
: The human mind is composed of sets of modules operating independently from each other.
Fodor
(1983): input module, but centralized, non-modular, system for higher-level processes.
Versus:
massive modularity hypothesis
Domain Specificity
Anatomical Modularity
: Each module is located in a specific and potentially identifiable area of the brain.
There is little support for the assumption of anatomical modularity.
Uniformity of structural architecture
across people.
Subtractivity:
brain damage can impair or delete brain functions, but cannot introduce new ones
Isomorphism
between the organization of the brain and the mind.
The study of brain-damaged patients can inform us about normal cognition.
Dissociation:
when a participant performs normally in one task but is impaired in a second one
<-- May occur because second task is more difficult...
Double Dissociation:
One patient performs normally on task X and is impaired on Y, second patient has impaired performance on task X, but performs normally on task Y.
Limitations:
Patients might have damage to more than one module.
A task might require more than one module
Not useful for demonstrating three or more separate systems.
Much more useful than
associations
between syndromes and symptoms.
Only
some
double-dissociations have theoretical relevance.
:red_cross:Limitations:
Impact of damage can be camouflaged by compensatory strategies.
In practice, brain damage is more extensive than just one module.
Seriality Assumption
seems to be incorrect.
"The seriality assumption states that cognitive processing is best described in terms of serial processing, modular information processing models of the sort encapsulated in “box diagrams.”"
There are sometimes large differences between patients with similar brain damage.
Cognitive Neuroscience
:star:
The study of the brain, as well as of behavior
Broadmann' s cytoarchitectonic map
Two major principles determine brain organization:
(Bullmore and Sporns, 2012)
Cost Control:
costs would be minimised if the brain consisted of limited,
short-distance
connections;
Trade-off between these perspectives: IQ not correlated with total number of brain network connections,
but
associations between IQ and global efficiency of functional brain networks.
Efficiency:
the ability to integrate information across the brain through numerous
long-distance
connections.
.
Neuroimaging Techniques
Single-Unit Recording
Microelectrode, single-cell extracellular potentials.
Hubel and Wiesel, primary visual cortex monkeys and cats
extraordinarily high spatial and temporal resolution
:no_entry: Highly invasive
Expensive
Higher cognitive activities involve larger populations of neurons.
Event-Related Potentials (ERPs)
EEGs, but presenting the
stimulus several times to produce event-related potentials (ERPs):
Limited spatial resolution, high temporal resolution.
Investigate the time course of processing.
Mainly of value when stimuli are simple and task involves basic processes occuring at a certain time after stimulus onset.
:no_entry: Requires many trials; skull and brain tissue distort electrical fields. Largely blind to subcortical activity.
Electroencephalogram (EEG)
Recordings of electrical activity measured at the scalp
Problem: background activity may obscure the impact of stimulus processing.
Positron Emission Tomography (PET) Scans
Detection of positrons from radioactively labelled water (glucose-based tracer) injected into the body
Reasonable spatial resplution, very poor spatial resolution.
Indirect measure of activity, invasive technique.
Magnetic Ressonance Imaging (MRI) and Functional MRI (fMRI)
Blood oxygen level-dependent contrast (BOLD)
Distortion in the local magnetic field
Better temporal AND spatial resolution than PET
MRI
tells only about structure, not function
More expensive than ERP
Magnetoencephalography (MEG)
Uses a super-conducting quantum interference device (SQUID) to measure magnetic fields produced by electrical brain activity. Magneric component of EEGs basically.
Excellent temporal and very good spatial resolution.
right thumb rule - direction of magnetic field induced by a current
:no_entry: Expensive, uncomfortable, technology and analysis are new and complicated.
Transcranial Magnetic Stimulation (TMS)
Limitations:
its effects are complex
it can only be applied to brain areas with no overlying muscle;
it may cause activity changes in brain areas distant from the stimulation site
there are possible safety issues (e.g., seizures).
Local transient disfunction of the brain that can be reversed after a couple of moments.
Reveals patterns of association between brain activation and behavior, rather than direct evidence of cognitive processing.
Most of it is based on the assumption of functional specialization. This may not hold true for higher-order cognitive functions involving greater integration & coordinationacross the brain.
Computational Cognitive Science
:star:
Computational modelling and programming computers to mimic aspects of human cognitive functioning.
Specifying a theory
and
predicting behavior in new situations
Explanatory and predictive basis for phenomena
Testing if a theory has hidden assumptions of vague terms.
Connectionist Networks
(aka: neural networks, parallel distributed processing PDP models): Neuron-like units (or nodes) connected together.
Units affect each other by excitation or inhibition.
Units take the weighted sum of all inputs and produce an output if a threshold is exceeded.
The network is characterised by the properties of the units, the way they are connected, and the rules
used to change the strength of connections
Networks can have different structures or layers
A representation of a concept can be stored in a distributed manner by a pattern of activation.
The network can store many patterns without them necessarily interfering with each other.
Many connectionist models based on the assumption that knowledge is represented in a distributed fashion.
Back-propagation of errors
: mechanism allowing a network to learn to associate a particular input pattern with a given pattern
by comparing actual responses with correct ones.
The model can be made to learn the behavior, rather than having the behavior explicitly programmed.
More Info on Connectionism
Production Systems
Useful characterizing processes such as problem-solving and reasoning.
IF/THEN
production rules.
Working memory
: matching the contents of working memory against the IF-parts of the rules and executing the THEN parts.