INT Lecture 4 - Theories of Intelligence

• Correlations form the basis of factor analysis
• Factor analysis groups variables that are most related, or correlated

Example Correlations

• A positive correlation is where as one variable increases, the other variable also increases, represented by a value of 1
• No correlation means there is no relationship between the variables, represented by a value of 0
• Negative Correlation is where as on variable increase, the other variables decreases, represented by a value of -1
• Height increases with weight as an example of positive correlation
• There is no correlation between height and hours spent gaming
• As family income goes up, infant mortality goes down, is an example of negative correlation

Psychometrics

• Psychometrics quantify psychological phenomenon
• Trying to capture unseen psychological processes using measurable items
• Measurable items include:
  • Behavioral observations
  • Questionnaires
  • Patient assessment
  • Test scores

Constructs

• To measure constructs like depression or intelligence, there is a need to find items that relate closely to these constructs, which is validity
• Items need to be consistent across samples and time
• If items reliably measure a process, such as depression or intelligence, they should all be related to each other
• We often want to create a single score for each person and trust that score is measuring something like high depression

Factor analysis

• Task is to group variables that are most related, or correlated
• Factor 1, is the hidden/latent variable made up of variables A, B and C which are observed variables

Personality Traits & Types Example

• Personality traits can be grouped into personality types.
• Personality traits:
  • Sociable
  • Hard-working
  • Warm-hearted
  • Outgoing
  • Dutiful
  • Helpful
• Personality types:
  • Conscientiousness
  • Extraversion
  • Agreeableness
• Questionnaires are one method to determine the factors
• After results are put into Excel, factor analysis can be run to determine if the traits correlate strongly with each other to determine if they can be clustered into a factor

Metric Variables Example

• First factor analysis on all 6 metric variables is calculated
  • Outgoing
  • Sociable
  • Hard-working
  • Dutiful
  • Warm-hearted
  • Helpful
• There is a need to choose a number of factors
• Factor analysis does not give a clear answer
• Choose the number of factors by sorting eigenvalues by size and plotting them on a graph
• Count how many eigenvalues are greater than 1
• In this example, you would select 2 factors

Eigenvalues

• For this example,.3 factors have an eigenvalue greater than 1, so we can select 3 factors

Analyzing Results

• Then look at the generated correlation matrix
• Factors can also be ranked by size according to explained variance and communalities, which shows how much variance is explained by 3 factors
• Rotated Component Matrix is used to find the loadings of the factors on the variables
• This rotation clusters the variables so that certain variables load as high as possible, and other variables load as less as possible on each factor
• Factor analysis does not tell us how to name the factors

Factor analysis

• Factor analysis provides a more complex method of defining an operational definition
• This method does not a use single variable, but rather several variables
• This is useful in defining the unobservable factor, like the g-factor or general cognitive ability

Mental Abilities

• A model for mental abilities includes the g-factor, which is common to all mental tests
• Spearman observed children's performance ratings across seemingly unrelated subjects were positively correlated, in 1904
• 1904 was the seminal paper on factor analysis of intelligence
• Spearman followed pioneering work by Galton using correlations
• Spearman developed the earliest version of factor analysis
• Spearman coined the term g or the g-factor
• Spearman developed his 2-factor theory of intelligence so ‘Under certain conditions the score of a person on a mental test can be divided into two factors, one of which is always the same in all tests, whereas the other varies from one test to another; the former is called the general factor or g, while the other is called the specific factor'

Spearman’s 2-factor theory of intelligence

• His theory included specific intelligences
  • Verbal
  • Mechanical
  • Spatial
  • Math
• As with general intelligence of (g)

Theory defined

• Intelligence is defined by underlying all the positive correlations
• G is the intelligence required for performance of intelligence tests of all types
• Mental energy underlies specific factors of intelligence
• Mathematical ability is not only affected by one's specific ability to perform a mathematical task but is also largely determined by general intelligence
• The positive correlation effect for above tests are consistently replicated

Unitary G Factor Challenge

• In 1938, Thurstone applied a different statistical method
  • This method involved weighing the observed variables differently
  • This yield not one but a set of “primary mental abilities"
  • Tests were positively correlated but scores on verbal tests were more highly correlated with other scores on other verbal tests than with scores on spatial tests
• Individuals have strengths and relative weaknesses in some cognitive domains
• This might suggest that there is no such thing as “g”
• Clusters of correlations suggest primary mental abilities
• (Not talking about specific abilities here)

Thurstone’s 7 primary mental abilities

• Thurstone suggested that there were seven separable human mental abilities
  • Word Fluency: Ability to use words quickly and fluency in performing such tasks as rhyming, solving anagrams, and doing crossword puzzles.
  • Verbal Comprehension: Ability to understand the meaning of words, concepts, and ideas.
  • Numerical Ability: Ability to use numbers to quickly compute answers to problems.
  • Spatial Visualization: Ability to visualize and manipulate patterns and forms in space.
  • Perceptual Speed: Ability to grasp perceptual details quickly and accurately and to determine similarities and differences between stimuli.
  • Memory: Ability to recall information such as lists or words, mathematical formulas, and definitions.
  • Inductive Reasoning: Ability to derive general rules and principles from the presented information.

Deemphasized G

• Thurstone de-emphasized 'g' and suggested instead that there were a number of stable and independent mental abilities
• These abilities were not general like g
• An example is a person might have weak verbal ability and strong numerical ability
• Each person possesses different levels of these seven factors
• These levels do not depend on one another, however the seven were positively correlated even in his dataset

Midlevel Abilities

• Emergence of mid-level abilities include:
  • Physical Coordination
  • Academic Skill.

20th Century Theories

• John Carroll book (1993) 800 pages long. Mammoth task of reanalysing over 400 sets of reliable data on human mental ability testing
• Studied:
  • 130,000 healthy adults
  • school children
  • infants
  • university students
  • people with learning disabilities and people with mental and physical illness
• His analysis was summarised in a diagram he called the ‘three stratum model' of human cognitive ability

Stratum Model

• ‘Three stratum model' of human cognitive ability includes:
  • Cognitive task
  • Cognitive domain
  • G

Conclusion

• People obtain a good score for at least four possible and non-exclusive reasons:
  • Good overall meaning have high general intelligence
  • Good at that sort of test meaning have high ability for that cognitive domain
  • Good at that specific test which means a have high ability for a specific cognitive task
  • Chance favored them on that particular testing day

Cattell

• General intelligence (g) can be subdivided into:
  • Fluid (gf) intelligence
  • Crystallized (gc) intelligence

Fluid intelligence (gf)

• Is primary reasoning ability
• Is abstract
• Is Not cultural
• Is Better tested by Raven's matrices
• An example is inductive reasoning in abstract sequences

Crystallized intelligence (gc)

• Is the acquired knowledge/skills
• Is related to cultural/historical background of individuals
• Is Better tested by Weschler scales
• An example is vocabulary and social rules

Psychometric vs Cognitive theories

• "Fluid intelligence refers to reasoning and novel problem-solving ability. It is distinct from crystallized intelligence, which refers to overlearned skills and static knowledge such as vocabulary. Empirically, Gf is strongly associated with working memory”
• “Crystallized intelligence is acquired knowledge and skills, such as factual knowledge. It is generally related to a person's stored information and to their cultural influences...Crystallized intelligence (for example knowledge) is intelligence that increases throughout life and is a reflection of one's cumulative learning experience.”

Methods

• Raven’s Matrices are better tests of Fluid intelligence
• Wechsler Tests are better tests of Crystallized intelligence

Approaches in intelligence testing

• Psychometric Approach:
  • Is Based on the findings of factor-analytic studies that have looked at the various psychometric properties of intelligence scales
  • Describes how people tend to differ but it cannot explain why
  • Focused on whether people answer items correctly
• Cognitive Approach:
  • Is based on another set of intelligence tests that highlight biological and physiological processes involved in intelligence
  • Focuses on how people answer the items, and why some people are better than others at various mental abilities
  • Is Underpinned by cognitive information-processing, and efficiency of problem-solving
  • Efficiency is partly dependent on the limits of memory storage and speed of processing

Cognitive Tasks

• Elementary cognitive tasks (ECTs) are simple tasks to measure cognitive processes
  • Includes understanding stimulus, stimulus discrimination, visual search, retrieval of information / perceptual speed, etc
• Measuring response time is used
• Measured by Median reaction time which is the average response time over a number of trials
• Meausred by Standard deviation of reaction time (RTSD) which is the individual variability in response time over a number of trials
• Meausred by Inspection time (IT) / Evoked potential (EP) which is the time people take to process visual or auditory information
• Elementary cognitive tasks (ECTs) are correlated with traditional measures of intelligence, the correlations averaging about r = .35
• Overall scores based on the use of various ECTs with different measures of response times are found to correlate with scores on general intelligence tests of between r = .50 and r = .70
• ECTs are useful because they involve no past learned information