4- Cognitive Theories Flashcards
1- What is g… positive manifold
IQ tests always show a
positive manifold : it is nearly
impossible to find a test that
doesn’t correlate with others.
Even if the tests are seemingly
quite different.
Is g a cognitive process
involved in all tests? Is it ..
- Raw mental power ?
- Speed?
- Working memory ?
- Executive functioning?
2- Simple sensory testing
Early interest in the simple sensory testing approach sprung from the work of Francis Galton in London, Wissler’s in the USA and Spearman 1904.
Looking for a simple sensory measure as an index of intelligence : strength of hand, visual acuity, cancellation of “A’s”, and reaction time
Premise = more intelligent ppl are more efficient at fine sensory discriminations
Looking for a correlation between simple sensory measures and IQ
3- Inspection time (IT)
- Inspection time
- Visual task that shows two different images. After a while, images will be partially masked, and you will be provided a prompt. Click the side of the screen with the image that matches the prompt.
- Assumed that task time assessed ”raw
mental power” - Had abnormally wide range of IQs which
inflated correlation - Sustained attention plays a factor- 2nd
attempts after fatigue were worse - Is just another indicator of processing speed
Early studies: correlations as high as -0.80 (Brand and Deary, 1982) or even .9 in Nettelbeck & Lally (1976)
Nettelbeck & Lally (1976) (and other studies) had ranges of up to 72 IQ points
in their study!
They included ppl w very low IQ
The wide range inflates the
size of the correlation.
* If the correlation is corrected for its
(exaggerated ) range, the estimate size of the relationship is r
= –0.50
Later meta-analysis : mean correlation
approx -0.30 or -0.5 if corrected for
restriction (Kranzler & Jensen, 1989; Grudnick & Kranzler, 2001)
Early studies, correlates really well with IQ
BUT overrepresented the extremes… so correlation is actually -0.3 to -0.5
Correlation strength depends on sample: if all the same, lower correlation. If extremes, bigger correlation
Inspection time correlates with IQ,
why?
Could it be differences in sustained
attention ?
Chaiken (1993) gave 2 IT tests: 1 early and 1
very late in a long session
Performance was worse on the 2nd
test
Performance on the 2nd test
correlates best with IQ
Interpretation ???
Suggests that sustained
attention was crucial
IT – IQ correlation reflects a factor of
vigilance ??
After close look, IT may not be
measuring “general mental
efficiency”
Correlation is driven by the
processing speed and perceptual
reasoning scales on the WAIS!
All timed tests, all tap into
speed
Correlation driven by Coding
subtest (formerly called “digit
symbol” WAIS-3)
Yet, Coding is actually not very
“loaded” in g : coding scores
do not correlate that well with
the overall “g”
Partial Conclusion
Inspection time reflect a
component of intelligence; speed.
It is NOT the central process behind
all tests that could explain the
positive manifold.
IT measures mostly speed of
processing, not “ the raw
mental power that explains g“,
as was initially thought .
Inspection time tests likely
measure the second level
factor called Gs.
IT = Gs IT ≠ g
4- Simple/Choice Reaction Time
Very dissimilar tests are highly
correlated and are loaded in g.
There must be some more
general function of the brain..
General Mental efficiency?
Perhaps speed of processing
could measure this underlying
variable?
The paradigm
1. Starting position
Rest finger on “home” button
2. RT (reaction time)
time elapsed from the moment a button
lights up and the moment the home
button is released.
Decision time
3. MT (movement time)
time elapsed since the home button is
released and the light up button is pressed
Execution of movement
The findings (Jensen)
Simple RT – without choices
Correlation RT & IQ ~ -0.2
Tiny increase in correlation with
number of alternatives
Correlation RT & IQ -0.18 to -0.23
- Was affected by the number of available button options
Hicks law (1952) :
Reaction time = LOG number of options (more options=longer reaction time)
BUT
Jensen’s initial data came
from University students
* Results skewed by university student sample
Restricted range of IQ ! = low correlation
Jensen and Munro (1979)
* RT (r = −.39) with Gf (raven)
* MT (r = −.43) with Gf (raven)
Detterman 1987 found a
correlation of .32
Deary 2001 found .29 (simple
RT) to .49 (4 choices)
*measures of dispertion (see image)
Reaction time (RT) correlates
with IQ, why?
1. differences in learning, or
practice effect?
RTs decrease with practice, in all
individuals
High IQ ppl show greater learning
ability – greater decrease in RT over
time, with practice
2. differences in sustained
attention in low vs high IQ?
Correlation between IQ and
variability of RT is about .2
Performance on sustained attention
have been shown to correlate with
RT (Carlson, 1983)
3. differences in attentional
lapses?
“worst performance rule” Coyle
(2003) – effect found in various studies
People’s longer reaction times rather
than their shorter reaction times
correlated best with IQ
* Lower IQ people generally had longer reaction time
Short reaction time correlate less
with IQ
Lower IQ individuals had more
instance long reaction time
Perhaps the long latencies
reflect trials when attention is
weak?
Partial conclusion : RT is too poorly
correlated with IQ to be an explanation of g
Theoretical problems:
1. Theoretically, the slope of RT is supposed to correlate with IQ
* But many of the other parameters
correlated as high or higher with IQ
(oups!).
aka * RT correlates with IQ just as much as other factors
2. The mean MT correlated with IQ
movement time correlates with IQ??
aka * Movement time correlates with IQ (when it shouldn’t)
3. Low reliability * good split-half reliability (above 0.66, with
a median of 0.84)
* low test-retest reliability (range: 0.39 to
0.84, with a median of 0.63) * RT slope has the lowest test-retest
reliability, of 0.39 (oups!) * Doesn’t work well with the assumption
that “intelligence as a stable trait”…
aka * Low task reliability-implicates IQ fluctuates frequently (which it doesn’t)
5- Processing speed - Salthouse
Reminder : Gf increases through adolescence, peaks in the early to mid-20s, and declines thereafter.
- Timothy Salthouse (1985, 1993, 1996) suggests:
the decline in fluid ability is the result of a
slowing of processing speed. - How does it work?
1. limited time mechanism: slower
processes occupy the available time,
higher level processes may not have
time to be executed
aka found longer
response time in older people
2. simultaneity mechanism: due to
slowing, the product of the first
processes may be lost when the later
processes are completed - This makes for 1 – longer response time and 2- less accuracy in aging
aka found less accuracy in older people
Shared variance
* Nearly 75% of age-related variance on
cognitive tests is shared with measures of
cognitive speed. (Salthouse1996)
His theory
there could be several speed factors at
work
* doesnt necessarily endorse that
decline in speed with aging is due to
a single factor (such as general
slowing of nerve conduction)
Ppl use different strategies
* “as individuals age, they adapt their
strategies on cognitive tasks to try to
compensate for the negative effects
of slower processing”
* These strategic choices can mask the
effects of aging in everyday tasks.
His legacy
1. Salthouse’s theory does not explain
intelligence per se, but he offers an
explanation for the decline in Gf in late
life.
2. He points out an important factor to
consider when designing education
and training for older adults: speed of
processing!
stimuli are encoded and analyzed more slowly
older individuals may not have simultaneous access to as many different pieces of information as younger learners would.
6- Working Memory
Cowan’s theory of intelligence (2014)
Simply put : the larger the WM, the more complex the ideas can be, the more complex the learnings can be Also, knowledge helps dealing with large amount of info in WM, reducing the WM load by chunking
familiarity with material will help span
Generally, IQ tests will include few or many of these:
* Digit Span forward (simple memory span)
* Digit Span backward (working memory span)
* Digit Span re-ordering (working memory span)
* Letter-Number sequencing (working memory span)
* Arithmetic (mathematical skills and working memory)
* N-back – not in WAIS (executive working memory)
Gf & WM
Substantial positive correlations between measures of WM capacity and IQ reported in many studies
WM spans show robust correlations with scholastic aptitude and Gf
Individuals with high Gf scores also provide faster reaction times and better WM scores
Some said WM span accounts for 50% the normal variability in Gf
The number of mental operations to be kept in WM correlates well with Raven’s and Gf.
Kyllonen and Christal’s (1990)
Interested in WM and reasoning
Ran 4 experiments
N = 2100 air force recruits
Paper pencil tests and computerised tests evaluating
1. Reasoning
2. Speed
3. General knowledge
4. Working memory
Stats : confirmatory factor analysis and structural equation modeling
Findings : The correlations between the reasoning factor and the
working memory factor in the 4 experiments were: 0.82, 0.88, 0.80, and 0.82
Conclusions : working memory plays a central role in intelligence
BUT
correlational research = we don’t know what causes what
Maybe WM scores allows better reasoning : greater
container for holding and manipulating more complex
thoughts ?
Maybe better reasoning allows greater WM scores : those
with high IQ are better at compensating - using their smaller WM more efficiently?
BUT
WM scores still correlates with Gf scores even when we take out the variance related to Digit Span Forward and Digit Span Backward (e.g. Kane and Engle 2003)
- They conclude its not “just the storage” that is important, it may be something else.
Maybe Executive control?
- Maybe some are just better at doing “controlled
tasks”, like resisting to distractions, for instance.
7- Executive control
The cocktail party phenomenon
Conway et al 2001
33% of subjects hear their
name in an unattended,
irrelevant message
Subjects who detect their
name in the irrelevant
message also had lower WM
Suggesting that
individuals with low span
WM also have difficulty
blocking out, or inhibiting,
distracting information.
IQ and EF, in short..
IQ and EF scores correlate frequently and significantly
Gc tasks not as strongly correlated with EF tasks .
Partial conclusions
g ≠ EF
Also..
EF ≠ unitary
* Highly correlated with IQ (fluid), but not as much crystallized
8- Creativity
Intelligence, EF and Creativity Benedek et al 2014
Fluid intelligence (Gf)
* Numerical-inductive
reasoning * discover the rules in a
number series
* Verbal-deductive
reasoning
* Syllogism tasks
Creative ability
* “Divergent tasks” : generation of
creative ideas to an open problem
(Kaufman, Plucker, & Baer, 2008)
* “Find creative alternative uses for a brick” * “What can be used for speedy travel?”. * Responses scored for creativity
Executive
Functions
* updating
* Inhibition
* shifting
(see image)
9- Hunt and the Cognitive Correlates Theory
Is an approach to understanding intelligence popular in Hunt’s lab
Participants are tested using ∑ cognitive psychology experimental paradigms to inform about basic cognitive abilities
Looking for correlations between tests . Goal = identify the cognitive components involved in each tests
For example, verbal ability would be related to the rapidity of
processes in short-term memory.
Also, quantitative ability would be related to resistance to interference in memory.
Generalizations are made from multiple findings over several experiments.
Hunt hoped this would bridge psychometric psychology and
cognitive psychology.
In other words:
* Attempting to bridge
psychometric and cognitive
theories
* Found that:
* Verbal ability related to processing
speed and short-term memory
* Quantitative ability related to
resistance to memory interference
Example: Hunt’s lab
Tested undergraduate students at University of Washington on a
wide range of cognitive psychology experimental paradigms
Selects well-know, recognized experimental tasks
Posner’s [Posner & Mitchell, 1967] letter match/name matchprocedure (shallow or deep processing)
Wickens’ [1970] release from proactive inhibition procedure Brown-Peterson [Peterson & Peterson, 1959] short-term
memory procedure (STM with resistance to distraction)
Atkinson and Shiffrin’s [1968, 1971] continuous pairedassociates procedure (complex attention)
etc
extreme ability groups method
Take student from top quartile vs. bottom quartile on X ability
to determine which process is crucial to which ability *Participants *top or bottom quartile of a college entrance test for verbal ability
*top or bottom quartile of a college entrance test for quantitative ability
*crossing of these 2 dimensions produced 4 groups *high verbal, high quantitative (N = 30), high verbal, low quantitative (N = 25)
* high quantitative, low verbal (N = 26), low verbal, low quantitative (N = 23)
*Hunts conclusions *First, verbal ability appeared to be related to the rapidity of processes in short term memory. *Second, quantitative ability appeared to be related to resistance to interference in memory. *These generalizations are based upon multiple findings over several experiments.
10- Sternberg and the cognitive components theory
Sternberg pioneered the “cognitive components”
approach to understanding intelligence (late 1970s and
early 1980s)
He investigated a typical task : the analogy.
His model had 7 processing components: encoding,
inference, mapping, application, justification,
comparison, and response.
Sternberg believed that much of what we termed
“intelligence” could be accounted for by information
processing components.
-Encoding was significantly
correlated with reasoning, but in
a positive direction (r = 0.63,
Sternberg, 1977).
-Response component was strongly correlated (negatively) with reasoning (r = –0.77)
-Inference and mapping = negative correlations (–0.54 to –0.56) with
Gf. but do not correlate well with speed of
perception (Sternberg 1977).
Indicates these component latencies are
NOT measuring simple speed of responding
Sternberg and Gardner work together (1983)
Extended Sternberg’s model (1977) on 2 other typical IQ
tasks: series completions, and classifications.
Includes many of the same information processes found in the
analogy model : they actually suggested that the same
information processes could underlie intelligent behavior
across a range of tasks.
Replicated Sternberg’s (1977) findings
Criticisms : lack of generality.
How is this explaining real world display of intelligence?
(outside the testing environment)
Questions like these led Sternberg to develop his “triarchic
theory of intelligence”
Summary:
* Used visual ‘analogies’ to measure
cognitive components
- Encoding, Interference, Mapping,
Application, Justification,
Comparison, Response execution
- (EIMAJCR) Eating Ice Makes a Juicy
Cold Refreshment
* Intelligence correlates with more
time spent on encoding, less time
on interference and application
* “Triarchic theory of intelligence
11- So what is g?
Seems like it’s just a statistical
entity
Not necessarily a single cognitive
process
But why do we find a general
correlation factor, across
all IQ batteries?
- Mutualism
- Associationism
12- Mutualism
Mutualism theory
Premise : “ cognitive processes have mutual beneficial or facilitating relations. “
Each process supports the development of other processes.
Some studies in development show that scores on different IQ subtests are more correlated in young children, then correlation becomes weaker, then it
becomes stronger again (e.g. Li et al, 2004)
In other words:
* Each cognitive process supports
the development of other
cognitive processes
* The relationship between
processes fluctuate over the
lifespan
13- Associationism
Associationist theory
Associationist theory of multiple processes
There are millions of processes involved in the WAIS battery
Tests that load high on “g” are simply tasks that sample a high proportion of these millions processes
Examples :
» Thorndike (1925)
» Thomson (1916, 1936)
» Hull (1928)
14- Summary of cognitive theories
Cognitive theories of intelligence, in short :
* Tried to understand intelligence in terms
of the cognitive processes that underly it.
* Tried to understand “g”
Example of cognitive theories of intelligence
suggested intelligence was …
1. elementary sensory and motor processes -
Galton and Catell
2. efficient visual processing - for IT
researchers such as Nettelbeck
3. ability to process increasingly greater
amounts of information and make a simple
decision - Jensen
4. working memory - Kyllonen and Christal
5. STM (for verbal ability) and the ability to
resist interference in memory (for
quantitative ability) - Earl Hunt and
colleagues,
6. processes involved in solving intelligence
test items such as analogies, series
completions, and classifications - Sternberg