Transcript Current Theoretical Approaches and Issues in

Current Theoretical Approaches and
Issues in Classical Conditioning
Psychology 3306
Everything you know is wrong
• So, the number of pairings is an important,
maybe all important, variable in determining
the amount of conditioning, right?
• Fine, then explain Kamin Blocking to me…
Kamin (1968)
Group
Control
Blocking
Phase 1 Phase 2
Nothing
L+
Test
Result
LT+
T
CR
LT+
T
No CR
L = Light T = Tone + = CS (shock)
Blocking is rocking
• Same number of tone shock
pairings in both groups
• It is NOT just number of
pairings
• The tone predicts nothing in
the blocking group (nothing
extra anyway)
• These results, and some
others, lead to the Rescorla
Wagner Model
You said there’d be no math!
• Yes, it is a math model
• Trial by trial
• Assumes you can get excitatory conditioning,
inhibitory conditioning or nothing
• All based on what the CS predicts
• Let’s look at the rules
The Rules
• If the strength of the US is greater than
expected then excitatory conditioning to the
CS is the result
• If the strength of the US is LESS than expected,
then you will get inhibitory conditioning
• The larger the discrepancy between what is
observed and what is expected, the greater
the conditioning
More rules
• The more salient the CS, the more
conditioning you will get
• Two or more CSs together, their strength is
additive
• This is, in essence, a model of surprise! The
more surprised the animal, the more it learns
The model makes some groovy predictions
•
•
•
•
•
Slope of the acquisition curve
Blocking
Conditioned inhibition
Overshadowing
Overexpectation
Overexpectation
Group
Phase I
Phase II Test
Result
Exp
L+ T+
LT+
L, T
Weak
CR
Control
L+ T+
nothing
L, T
Strong
CR
The Model:
•
•
•
•
•
•
•
Δvi = Si(Aj-Vsum)
i = CS
j = US
S = Salience
A = Value of the US
V = amount of conditioning
These quantities are, of course, hypothetical
An example
• OK, say a food pellet = 100
• Say salience of a light CS = .2
• Vsum = 0 (at the start of the experiment, there
is no conditioning yet
OK, now for the numbers
• Trial 1
• Δvi = Si(Aj-Vsum)
– =.2(100 – 0)
– =20
• Trial 2
– ΔVi = .2(100-20)
– =16
Continued….
• Trial 3
– Δvi = Si(Aj-Vsum)
– ΔVi = .2(100-36)
– =12.8
• And so on….
• Less and less conditioning as time goes by
• Cool eh
Overshadowing
•
•
•
•
CS1 -> Light, S = .2
CS2 -> Noise, S= .5
2 CSs, so two calculations per trial
Trial 1
– ΔVLight = .2(100-0) = 20
– ΔVNoise = .5(100-0) = 50
Overshadowing
• Trial
– ΔVLight = .2(100-70) = 6
– ΔVNoise = .5(100-70) = 15
•
•
•
•
•
OK, how does blocking work?
Well there is no strength left
Conditioned inhibition?
Negative for old CS
Additive model
Stuff it cannot deal with
• CS preexposure
• Change S?
• Mackintosh’s attentional theory does this, S
becomes an attention parameter
• Pearce Hall model
• Gallistel’s model
Types of associations
•
•
•
•
•
•
First order conditioning is S-S
Second order is S-S and S-R
CS - context associations too
US context associations
Context Blocking
CS CS associations in compound stimulus
experiments
• Occasion setting (Holland)
Constraints on Pavlovian Conditioning
•
•
•
•
Taste aversions
Not just sickness
Not the aftertaste
Only to certain elements of the food, which
depends on the species
• Special?
• Could just be a quantitative difference
(Andrews and Braverman, 1975)
Form of the CR
• CR is often like the UR but not always
– Weaker
– Opposite direction
• Drug tolerance
• Compensatory CRs with opiates
• Context as CS
– Shooting gallery effect
• Could depend on drug action being in PNS or
CNS (Stewart et al)
Physiological Basis
• New synapses formed in Aplysia
• Increase in transmitter release in neurons
sensitive to CS (very cool)
– Just like habituation!
• What about more complex creatures
Five points about Physiology and
conditioning
•
•
•
•
•
•
1) CR and UR pathways are often different
2) CR production is distributed
3) Conditioning is distributed
4) Different CRs, different brain regions
5) Sometimes it is individual neurons
My conclusion then is that we have a very
basic mechanism at work here