Conditioning and Learning

 

I.                 Some broad issues in learning

 

A.       A simple definition: Learning is a change in behavior resulting from experience; in evolutionary terms, learning is an adaptive change in behavior that results from experience

 

B.        The difference between maturation and learning: Some behavior change (walking, talking, adult sexual behavior) requires biological development as well as experience

 

C.        Simple vs. complex kinds of learning

 

1.       Relatively simple forms of learning: habituation, classical conditioning, operant conditioning

 

2.       More complex kinds of learning: learning to talk, learning calculus, learning the history of the Civil War

 


II.             Classical conditioning

 

A.  Pavlov’s dogs:  Pavlov originally studied the physiology of salivation, for which he won the Nobel Prize.  In the course of this research, he became aware of a kind of learning, which today is called, “classical conditioning.”   Sometimes, it is also referred to a “Pavlovian conditioning”

 

B.       The basic paradigm of classical conditioning:  A formerly neutral stimulus (the conditioned stimulus; a bell, for example) is paired with another stimulus (the unconditioned stimulus; food, for example) that automatically produces a response (the conditioned response; for example, salivation).  After repeated pairing, the neutral stimulus (the bell) will elicit a response similar to the unconditioned response (i.e., the bell will produce a response of salivation; this learned response to the bell is called the conditioned response)

 

Abbreviations:

US = unconditioned stimulus

CS = conditioned stimulus

UR = unconditioned response

CR = conditioned response      


C.       Some examples of classical conditioning:

1)    Learning to feel upset at the sight of flashing police lights in your rearview mirror

2)    Learning to feel anxiety when you hear the sounds at the dentist’s office

3)    Learning sexual arousal to objects that have been associated with sexual arousal in the past (e.g., items of clothing)

4)    Feeling tender emotions when you hear a song that was associated with your first romance

5)    A new mother whose breasts start to produce milk when she hears her baby’s cry

6)    Learning to feel emotional arousal to certain words (4-letter words, bigoted labels)

7)    The famous case of “little Albert” – learning fear

 

D.       Traditionally, psychologists believed that responses that can be classically conditioned are involuntary responses (examples: heart rate changes, gastric motility, sweating, eye blinks, sexual arousal).  This is in contrast to operant conditioning, in which voluntary responses are molded through their rewarding and punishing consequences

 

E.        What is the evolutionary “purpose” of classical  conditioning?  One answer: It helps the body prepare itself for an expected or likely event.  For example, if food is likely, salivation aids the digestive process.  If a painful shock is likely, the body prepares itself for this stressor.

 

F.         Some important terms and concepts in CC:

 

1.       Extinction: A weakening of the conditioned response when there ceases to be a pairing between the CS and the US

 

2.       Spontaneous recovery: The tendency for a conditioned response to reappear after extinction takes place

 

3.       Generalization: The tendency for an animal or person not only to condition to the exact CS used during conditioning trials, but also to similar stimuli; for example, if a dog is conditioned to salivate to a particular bell, it may also salivate to other bells as well

 

4.       Semantic generalization: a kind of generalization which occurs only in people;  when people learn conditioned responses to words, they may generalize the responses to the objects or concepts that the words refer to.  For example, if you learn prejudiced feelings to a bigoted label, you may generalize them to the people referred to by the label;  Also, when people learned conditioned responses to words, they may generalize the responses to words with similar meanings.  For example, if the word “white” is paired with electric shocks and you learn to be afraid of the words, you may also show fear responses to the word “light,” which is semantically related to the word “white”

 

G.       Some factors that influence classical conditioning

1.       Time delay between CS and US: Usually conditioning is strongest if the delay is between 250 to 700 milliseconds

2.       Time arrangements of CS and US

 

a.       forward, trace, simultaneous, and backward conditioning:

forward – CS comes first, and while it’s still going, the US occurs

trace – CS comes first, and after it stops, the US occurs

simultaneous – CS and US occur at the same time

backward – CS occurs after the US has started

 

H.       The contiguity vs. contingency views of classical conditioning

 

1.       contiguity (Pavlov’s view):  CC occurs when the CS and US occur together, in time and space; (Note similarity to British associationist views)

2.       contingency: CC occurs only when the CS provides some information ahead of time about the likelihood of the US occurring

 

 

3.       Some evidence:

a.       effects of CS, US time arrangements

b.      Blocking experiments – What happens if animals are first conditioned to blink (CR) to one CS (a sound) for 8 trials, and then a light and a sound (two CS’s) are paired with the air burst for another 8 trials?  Will the animal show a CR to the newly added CS?

 

I.            Is classical conditioning a kind of stimulus-response learning (the CR is “hooked to” the CS), or is it a kind of stimulus-stimulus learning (the animal learns that the CS “signals” the US)?

 

1.       the “response-prevention paradigm” – What happens if we prevent an animal from making the CR by paralyzing the muscle?  When the paralysis is removed, will the animal show the CR?


2.       the “US devaluation” paradigm – In Pavlov’s original experiments, the dogs were hungry.  What happens if we condition dogs to salivate to a bell, and then allow the dogs to eat until they’re stuffed.  Will they then salivate to the bell? 

 

 

J.           Biological preparedness and classical conditioning

 

1.       the unusual case of learned taste aversions:  conditioning can occur in on trial; the time delay between CS and US can be long

 

2.       animals can learn some kinds of CR (food aversions) more readily to some kinds of CS (smell, taste) than to others (visual cues)


III.         Operant conditioning

 

A.       Edward Thorndike’s (1898) cat puzzle box: Hungry cats locked in a box, which could be opened only if the cats pulled an unlatching device (a loop of wire); at first cats randomly moved, meowed, and clawed, but gradually they became better (quicker) at getting out of the box with successive trials

 

B.        Thorndike’s “law of effect”: rewards or reinforcers strengthen stimulus-response connections; a mechanistic, unthinking view of the effects of reward

 

C.        Is reward necessary for learning to take place?  Tolman’s notion of “latent learning” – a rat allowed to freely roam through a maze stills seems to learn it layout, even when the rat is not rewarded

 

D.       B. F. Skinner’s (1904-1990) view of operant (or instrumental) conditioning

 

1.       Animals emit behaviors freely, called “operants”; for example, rats in a “Skinner box” might press a lever sticking out of the wall of the box; a reinforcer is anything that increases the probability of a response when it follows the response (Examples: food is reinforcing to a hungry animal; water is reinforcing to a thirsty animal; sex can be reinforcing to a sexually mature animal)

 

E.        Some important concepts and terms in operant conditioning:

 

1.       Positive vs. negative reinforcement: Both increase the probability of a response; however, positive reinforcement is the presentation of a desired stimulus (food, money), whereas negative reinforcement is the termination of an aversive or unpleasant stimulus (electric shock, pain, anxiety).

  

Example to think about: Do people drink alcohol or take drugs like cocaine because of positive or negative reinforcement?

                    

2.       Primary vs. secondary reinforcers:  Primary reinforcers are unlearned and ‘wired in” to the organism (example: food, water, sex); Secondary reinforcers are learned reinforcers (examples: money, school grades, tokens that monkeys work for to get treats)

3.       Schedules of reinforcement: interval (based on time) or ratio (based on number of responses); fixed (occurring after set intervals of times or fixed numbers of responses) or variable (occurring after variable time intervals or a variable number of responses); different schedules of reinforcement produce different patterns of response in animals and people

4.       Partial (some of the time) vs. continuous (every response reinforced) schedules of reinforcement; The partial reinforcement effect:  Partial reinforcement produces responses that are more resistant to extinction

 

5.       Punishment: an aversive stimulus delivered after a behavior which decreases the probability of (or even eliminates) a response

 

Skinner’s early view of punishment:  It is often ineffective because it may lead only to a temporary, situation-specific suppression of a response; furthermore, punishment is at best a partial strategy: it may eliminate an undesired response, but it doesn’t necessarily establish a desired response in its place; also, physical punishment may produce anger and modeling of aggressive behavior

 

When is punishment effective?

 

1)    In animals, punishment must be delivered soon after a response to be most effective

2)    Punishment must be strong (as severe as is ethically or practically acceptable) to be effective

3)    Punishment must be delivered consistently (compare this with the partial reinforcement effect)

4)    Punishment should start out strong; it should not start out weak and build up with repeated “infractions”

5)    Punishment is less effective if animal earlier experiences random and noncontingent punishment  (example: a child is randomly abused, and then punished for a specific “bad” behavior)

6)    Punishment is more effective if animal if offered an alternative response to the punished response

 

IV.       Language and conditioning: Can language learning be explained using classical and operant conditioning?

 

A.       Classical conditioning of emotional reactions to words

B.        Operant conditioning of early verbal utterances: Are some words reinforced?

C.        Limitations to the conditioning approach: Chomsky’s (1959) critique of conditioning approaches     

 

Language and Thought

 

I.                 Communication in infra-human species and characteristics of human language

 

A.Some examples of communication in lower animals

 

1.       the dance of the honeybee

2.       communication in jackdaws (a European blackbird)

3.       communication in apes and attempts to teach chimps and gorillas to “talk”

 

D.       Some characteristics of human language

1.       Natural human languages are based on relatively small sets of speech sounds called phonemes

2.       Syntax or grammar: All human languages have complex structural rules; Note, “grammar” here does not refer to the “good grammar” you learn in school

3.       Use of arbitrary, non-representative meaningful symbols: words and morphemes

4.       Generativity: All human languages can generate an infinite number of meaningful statements or sentences

5.       Learning: All human languages are learned; however, the human capacity for language and language learning may have a strong biological basis

6.       Cultural transmission of information: Human languages permit the storage and transmission of complex cultural information from generation to generation

 

E.        Language and thought

 

1.       Does thought require language? -- the case of Helen Keller

2.       The Whorfian hypothesis (also known as the Sapir-Whorf hypothesis or the linguistic relativity hypothesis): the language that people speak influences the way they think

a.       Strong vs. weak form

b.      Lexical vs. grammatical form

3.       Some psychological research relevant to the linguistic relatively hypothesis

a.       color labels and color perception in various cultures

b.      linguistic labels and their effects on memory

  

 

Intelligence

 

I.                 Early attempts to study intelligence

 

A.    Galton (a cousin of Charles Darwin) and his book, Hereditary Genius; tried to use simple measures (like grip strength, pain sensitivity, memory for dictated consonants) to assess “intelligence”

 

B.    In the United States, Cattell (1901) conducted a study which suggested little relationship between simple sensory measures and Columbia college students’ test scores

 

C.    Modern research has returned, using more sophisticate methods, to the question of reaction time and intelligence. Note, common sense views relate speed of thinking and intelligence, as is illustrated by phrases such as “quick witted” and “slow minded”

 

II.             Binet’s seminal work on intelligence

 

A.    A practical problem: The Paris school system sought Binet’s help in “objectively” indentifying “dull” school children

B.    Binet was pragmatic and he tried many different methods to measure “intelligence,” including digit recall, measuring size of cranium, assessing moral judgments, graphology, and even palmistry

C.    Defining intelligence: Binet eventually came to see intelligence as involving many processes related to 1) “the tendency to take and maintain a definite direction” in thought, 2) the capacity “to make adaptations,” and 3) the power of “auto-criticism”;  In simple terms, according to Binet, intelligence involves purposeful, directed thought, which successfully achieves goals and which is self-critical and self-correcting;  according to Binet and Simon (1916), the core of intelligence is: “judgment, otherwise called good sense, practical sense, initiative, the faculty of adapting one’s self to circumstances. …to judge well, to comprehend well, to reason well….”    

D.    Modern conceptions of intelligence hold on to Binet’s notions; for example, Sternberg’s triarchic (three-part) model of intelligence describes three aspects: 1) analytic intelligence (“school smarts”; the ability to analyze and solve academic problems), 2) practical intelligence (“street smarts”; the ability to apply knowledge and solve problems and achieve success in everyday life), and 3) creativity (the ability to invent, to come up with new solutions and views; divergent thought).

E.     The first intelligence tests:  Binet-Simon scale was published in 1905; revisions were published in 1908 and 1911; “difficulty” or “level” of questions (items) was determined by age-related changes in performance

F.      The notion of “general intelligence”: Binet observed that children’s performance on various questions tended to be positively correlated; that is, children who tended to do well on one set of items also tended to do well on other items, whereas children who tend to do poorly on one set, tended also to do poorly on other items

G.    Stanford-Binet:  American version of the Binet test; these are individual, not group tests

H.    Modern group IQ tests, such as the Wechsler

 

III.         Characteristics of good IQ tests

A.    Reliability: Is the test measuring something consistently; There are various kinds of reliability: test-retest reliability and internal consistency are two important kinds

1.  In general, IQ tests are quite reliable; for example, test-retest reliabilities are on the order of .9

B.    Validity: Is the test measuring what it’s supposed to measure? Do test scores predict what you would expect them to predict?  In the case of IQ scores, you might expect that they would predict school grades, job success, and intellectual accomplishments (e.g., scientific discoveries, patents, creative accomplishments such as publishing books)

1.       Cox’s estimates of the IQ’s of famous people like Mozart, Goethe, and Mill

2.       Terman’s “gifted children” study

3.       Research on IQ and school grades

4.       Research on IQ and job success

 

IV.       Nature, nurture, and intelligence

A.    The concept of “heritability”: Heritability is the computed proportion of population variability in a trait (e.g., variability in height or in intelligence) that is due to genetic factors; Note, heritability only applies to populations, not to individuals

B.    Ways of assessing heritability: Behavior genetic studies look at patterns of trait correlations in identical and fraternal twins and in families with adopted children.  Two clear examples: Trait correlations between identical twins reared apart, and trait correlations between adopted children and members of their genetically unrelated adopted family

C.    Behavior genetic studies often investigate two kinds of environmental influences: 1) Common environmental influences, which affect all children in a family the same way, and 2) Unique environmental influences, which affect various children in a family differently.  In common-sense terms, common environmental influences tend to make siblings more similar to one another, and unique environmental influences tend to make sibling different

D.    Behavior genetic evidence on intelligence:

1.       Heritability of IQ tend to be the range of .5 to .80 (i.e., 50% to 80% of the variation in adults’ IQ scores is due to genetic factors); heritability is lower for children and higher for adults

2.       Common environmental effects in children account for about 20-25% of the variability of children’s IQ scores, but 0% of adults

3.       Thus, in adults most of the non-genetic variability in IQ seems to be due to unique environmental effects

E.  Behavior genetic statistics on IQ do not give us useful information about group differences in IQ