Transcript Socio-Bio-Cognitive Learning ppt.

The Cognitive
Rebuttal
sociobiological
theories
of learning
Do Now
List (any) potential
problems with the
behaviorist
approach
The cognitive approach revolves around
concept-learning
• Behaviorism = black box
– S=>R learning
• Automatic associations
– Need experience to make
connections
• EX: Rat’s in Thorndike
puzzle box
– Connections based on
physical markers, not
meaning
Cognitive = concepts
– S=> S between mental concepts
• Access to mental processes
• Associations b/w mental
representation of CS and UCS
in classical conditioning
– Concept can be applied w/o
experience
• Demonstrated by pigeons
ability to learn a schema for
“tree”
S=>S Learning
• As opposed to seeing classical conditioning as physical
processes, they can be reframed as mental and
representational process wherein the bell brings to mind a
representation of food (bell => mental representation =>
salvation).
– This approach was denied by behaviorists because of its emphasis
on observational causes and effects
• Sà S theories can account for second order conditioning,
whereby a rat is conditioned to a light and then the light is paired
with a sound. The rat continues to respond to the sound before
extinction.
• The loud sound was habituated through trials for half the rats
• S => R = Rats should continue to freeze because now there is a
connection between the lights and freezing
• S => S = Rats should not freeze because they are connecting
light and sound
S=>S Learning
People and Animals show conceptual learning
•
•
•
Interpretation more important than physical characteristics
– Lemon squeeze paired with style, urn, freeze and serf
• Activated response with fashion, vase, chill and wave
• Not activated stile, earn, frieze, and serf
Meaning can be generalized
– Soviet boy conditioned to salivate to “good” but not to “bad”
• Salivates to “The soviet army was victorious” but not to
negative statements, such as “The pupil was rude to the
teacher”
Discrimination learning in pigeons shows conceptual learning
– Pigeons conditioned to respond to set of 80 sides that contained
either trees or part of trees
– Pigeons would respond equally well without shaping to new tree
slides, showing that a concept was derived of a “tree”
Cognitive Classical Conditioning
• Expectancy theory
– Not S=>R (automatic)
– Instead CS causes animal to expect some stimulus (mental
representation) and then a physiological response follows
that is seemingly reflexive to the stimulus
• Predictive value
– Effectiveness of classical conditioning dependent on
predictive value of CS
• CS must precede unconditioned US
– If not actively seeking knowledge, no reason to pay
attention to CS
• CS must signal heightened probability
– Internal mathematics to determine whether or not to
respond to CS
– Already having a good predictor prevents
conditioning
Cognitive operant conditioning
• Behaviorist perspective:
– Reinforcer important only in that it strengthens connection
between S and R
• Cognitive perspective
– Added complexity (or complexity not acknowledged by
behaviorists)
– S => R the stimuli in the box => lever press
– S => S the discriminative stimuli => mental image of food
– R => S pressing the lever => mental image of food
Cognitive Operants
•
Means-end relationships
– A tone leads to two ideas
• Knowledge that pressing lever brings food
• Decision to press lever, depending on interest
– Rats conditioned on either dry food or sugar water and then deprived of
food and water
• Rats conditioned on sugar water would press in high rates (satisfied both
thirst and hunger)
• Rats conditioned on pellets did not press as often as if rats understood
pellets would not alleviate thirst and hunger
• These results do not imply conscious decision making, but rather that
some mental process is taking place
Cognitive Operants
•
•
•
Reward contrast effects
– Different response rates if the attractiveness of the reward is changed
• Reward devaluation
– “this will surprise few people. What might be more surprising is
that the issue was not properly settled long ago.” - Anthony
Dickinson
– Negative contrast effect: decreased reward attractiveness, drop in response
rate
– Positive contrast effect: increased reward attractiveness, increase in
response rate
Over-Justification Effect:
– Cannot be explained simply by association learning
Delayed gratification
– Able to consider rewards into the future
• pigeons even capable of such thinking
Place Learning
•
•
Behaviorist view
– Place learning is learning a
sequence of S=>R
• Expect subjects to only
know location if follow
exact route
Cognitive view (Tolman)
– Access to mental map
• Manipulation and
application of knowledge
– When barriers were removed
from the maze, rats used the
quicker route even though
they were never trained on
that path
• Shows latent learning
– Manifest only when
reward present
Observational Learning (Bandura)
- Kids who see adult beat on Bobo more
likely to beat on Bobo
- Kids who see adult rewarded for
beating on Bobo more likely to beat
on Bobo than other kids (vicarious
reinforcement)
- Kids who see adult punished for
beating on Bobo less likely to beat on
Bobo than other kids (vicarious
punishment)
Impossible to learn everything through direct experience
Four conditions for observational learning
•
•
•
•
Attention (must notice behavior, get better results if kids encouraged or rewarded for paying
attention)
Retention (need to remember situation to imitate it)
Motivation (this is where vicarious punishments and rewards come in)
Production (can't imitate something you are unable to do)
•
Animals also show observational learning
–
–
Mechanism for teaching offspring
Monkeys can learn how to crack nuts by observing other monkeys
Play as a Learning Mechanism
•
•
•
•
•
Play occurs in a context in which
behavior might be seen as
irresponsible is now allowed or
sanctioned
The drive state is absent from the
behavior (e.g., fighting without
anger or preying without hunger)
Play usually involved a high degree
of repetition
Play behaviors occur in sequences
that would not be expected in
serious contexts
There are clear signals that animals
intend to play
Evolutionary Perspectives on Learning
•
•
•
•
Ecological Perspective
– Context key
– Learning occurs in and as a result of the environment
Species-specific Fear-related learning
– Humans more likely to acquire conditioning for fearing snakes than
cucumbers
• Rhesus monkeys not naturally afraid of snakes but quire to
acquire
• Could be passed down through videotape through observational
learning so that rhesus even began to fear toy snakes
Food Aversion Learning
– Evolutionarily critical to quick acquire and react to information
pertaining to what to eat
Food-Preference learning
• Rats preferred tasteless, odorless substance (thiamine)
– Rats would eat two foods at a time as if to figure out
nutritional value, even when the thiamine as added to
different foods (not just a preference for a single flavor)
Food Aversion
Learning
•
•
Classical Conditioning
interpretation
– Sickness = withdrawal
(don’t eat)
– Sickness + food =
withdrawal (don’t eat)
– Food = withdrawal (don’t
eat)
Cognitive interpretation
– Effect evident even if sickness follows 24 hours later
• Classical conditioning dependent on time cooccurrence
– Distinctiveness of taste/smell critical
• Suggests accessing concept of food, rather than just
responding
– Preference for nutritional content over long term
– Observational learning for food preferences