CH 4: CLASSICAL CONDITIONING: MECHANISMS

Conditioning depends on:

• Prior experience with each stimuli (CS and US)

• How relevant CS and US are to each other

• Presence of other stimuli during conditioning trials

• Not always reliant on CS-US pairing 

Need to know the baseline responses to the CS and US before you begin an experiment:

• same stimulus may serve as a CS in some experimental situations, and a US in others 

  • e.g., Using food as CS in CTA, but use food as US in sign tracking experiment

• Prior experiences will have an impact on the current situation 

Novelty of stimuli

• Prior experiences with stimuli affects future learning 

  • Stimuli is new in the following cases 

Latent-inhibition (CS-preexposure) effect:

• Caused by repeated exposures to the CS before the CS is used in conditioning trials

  • Already have knowledge of CS not being relevant 

  • Learning is slowed

US-preexposure effect

• Caused by repeated exposures to the US before the US is used in conditioning trials

  • Already have knowledge of the US and it is positive

  • Learning is quickened

Salience of stimuli

• The stimuli to be conditioned must be noticeable 

  • Intensity; more attention-grabbing 

  • Biological relevance to animal’s environment 

  • Biological relevance to animal’s needs

• More naturalistic CS cues = More Salient

CS-US relevance (belongingness)

• Do the CS and US “go together” naturally? 

Garcia + Keolling (1966) study: relevance example

• Rats are separated into 2 separate groups

• Some receive shock (physical), some made ill (internal)

• Both groups drink bright noisy tasty water (bright light, loud sounds, and sweet water)

• Goal: find which component of the water naturally is linked to the states of discomfort?

• Group with sickness → went for audiovisual (sound and noise) bottle

  • Already got sick → assume it is bc of the taste component 

• Group with shock → went for tasty water (sugar water) 

  • Already shocked → assume it is bc of external stimuli

Learning without an US

• Pavlovian conditioning is limited if direct exposure to US is required for learning to occur

• Pavlovian conditioning can also occur in situations without a US

• 2 forms:

1. Higher order conditioning

2. Sensory preconditioning

Higher order conditioning

• Further you get away from association → weaker the response 

• CS may serve as US, once conditioned 

• CS₁ and CS₂ 

• CS₁  goes through conditioning WITH US → CS₁ becomes US when faced with CS₂

Sensory preconditioning 

• Allows us to have 2 CS

• CS₂ is never paired with a US

• What is applied to CS₁ then eventually applies to CS₂

CS₁ triggers reaction, CS₂ is close enough to CS₁ → CS₂ provides same reaction as CS₁

Stimulus-substitution model

• CS activates neural circuits previously activated by US

• If CS is coming in to activate neural pathway between US and UR → then US and UR change

  • CS becomes a surrogate/substitute US

  • * changing US will drag the CS bc there is only one pathway between US and UR 

• Treat the CS like a US

What makes a CR?

• US as a determining factor for the CR

  • Response is dependant on the US that elicits response

Role of the US on the form of the CR (jenkins + moore)

• Subjects: pigeons( within group design) 

• Autoshaping: actual door key illuminated → delivery of the US

  • Turns key into CS 

• Pigeons trained alternatively with food then water (vice versa) as US

• CR_food: pecked response key as if eating; rapid pecks with beak slightly open

  • CR produced due to the instinctive response of the US (food) 

• CR_water: pecked response key as if drinking; slower pecking with beak closed and slight tilt back, often with filtering water and swallowing

  • CR produced due to the instinctive response of the US (water)

CS as determining factor of the CR

• Similar to other sign tracking experiment above → rat used at CS to signal food

• US Predictions based on stimulus-substitution model

  • Ex. biting as if eating food

• However, CS elicited social affiliated CRs → might be part of some biological process 

• Does not support model that explains the CR only in terms of US

• This demonstrated that CS also helps determine the CR

  • If lever is CS for food, wat WILL gnaw or bite it f

CS-US interval as determining factor

• Short vs long CS-US intervals

  • Short intervals may not always prepare beforehand but longer intervals will allow for better preparation of actions/events

Short vs long CS-US interval example: sexual conditioning in mail quail

• Conditioned male quail with 2 CS-US intervals (1 min warning or 20 min warnign)

  • Also had unpaired control groups (helps us make sure its not pseudoconditioning) 

• US was accessible to receptive female at end of time interval

• Measured the type of CR in the 2 delayed conditioning procedures 

Determining factors for CR

• US, CS and CS-US interval all affect conditioned responding:

  • US - Jenkins & Moore’s pigeons (1973) 

  • CS - Timberlake & Grant’s rats (1975) 

  • CS-US interval - Akins’ quail (2000)

• We must consider how pavlovian conditioning functions in natural history of organisms:

  • Behaviour systems theory

S-R learning

• Instead of having previous pathway to build up on → makes new pathway instead

• What ends up being learned: CS produces CR through new pathway

  • US is no longer relevant

S-S learning

• Stimulus-stimulus 

• CS activated a representation of the US

  • In line with pavlov stimulus substitution

US devaluation paradigm

• US becomes less meaningful

• Ex. satiation of food → dog will not respond to CS bc the US is meaningless now 

• 3 phases

• if S-R, both groups should respond at high levels to CS, as during phase 1

• If S-S subjects in experimental group (where US was devalued), then subjects should respond at lower level to CS compared to control 

Phase 1

• Regular conditioning

Phase 2

• US magnitude becomes smaller

Phase 3/Test

• CS produces smaller response bc of the small magnitude US 

Holland + rescorla

• Rats responded less (lower CR) to presentation of the CS following US devaluation

• Supports S-S learning

• However, it's also empirical evidence that subjects learn direct S-R associations 

Conditioned diminution of UR

• Reduction in magnitude of response to an US caused by presentation of a CS that has been conditioned with that 

• Pavlovian conditioning modifies responding to US

Pavlovian conditioning modifies responding to US example

• Conditioned drug tolerance example

  • CS = cues related to drug-taking (e.g., location, paraphernalia) 

  • US = pharmacological/chemical stimulation of the drug 

  • UR = physiological effects of drug (e.g., analgesia in morphine) 

  • CR = physiological effects (same or opposite of drug) following exposure to strongly-associated CS

  • Key prediction: when there is no CS → drug tolerance will be reduced bc learned responses are slow being forgotten 

Blocking effect:

• Interference of the conditioning of NS bc of the presence of a previous CS

• Why is it a big deal?

  • Until blocking, temporal contiguity was considered to be sufficient for learning associations

• Nothing was actually learned ab new stimulus bc we knew enough about the previous CS

  • A blocks B from being learned 

Why does A block B?

• Kamin proposed that US needs to be surprising to be effective in producing learning (never seen before) 

• If US is reliably signaled by previously conditioned stimulus (A), won’t be surprising (nothing new to be learned)

  • No learning about B

  • No preparing for the US 

• Effectiveness of US + its surprisingness = basis of Rescorla-Wagner Model

MODELS OF ASSOCIATIVE LEARNING

• Rescorla-wagner model

• Attentional models

• Relative-waiting-time hypothesis

• Comparator hypothesis 

Rescorla-Wagner Model formula

∆ 𝑉 = 𝑘 (𝜆 − 𝑉)

∆ 𝑉

• trial-by-trial change in associative learning strength of US (i.e., learning) 

k

• constant related to salience of US (always a constant → more salience = better learning)

 (𝜆 − 𝑉)

• can understand as the amount of surprise

  • Mainly looking at what's inside the bracket 

𝜆 (Lambda)

• Maximum possible associative strength of US 

  • (what occurs)

  • Ex. you get food when you didn't expect it

𝑉

• Current associative strength of US 

  • (what is expected/what we know)

  • Ex. you don't expect food when you did get food

• Surprising = different from what is expected

R-W and blocking 

• Because B is being blocked by A → there is no surprise factor

• Example equation:

  • (𝜆 − 𝑉_(A+B))

  • If 𝜆 is V_a → and V_a is 100

  • If b is blocked → meaning no outcome = then b is 0

  • That means

  • (100 - (100-0)) = 0 

• Therefore: there is still no surprise BECAUSE b is being blocked by A

  • We can understand this as no learning from B, because we already know everything from A

R-W + conditioned inhibition

• R-W allows stimuli to only have one value, (+) or (-), not both

  • Inhibitory or excitatory

• Need to consider reinforced and unreinforced trials separately 

• Need to consider as multi-equation system

R-W as a multi-equation system

• CS+ means that US is coming

• CS+ and CS- means that no US is coming 

• Associative value of CS+ is lambda 

• CS- must then have value of negative lambda 

• CS+ acquired excitatory properties first

  • If CS+ and CS- together → equal in magnitude

Extinction of conditioned inhibition

• Note: R-W of extinction is problematic 

• CS- alone predicted to lead to loss of inhibition

• Model views extinction as reverse of acquisition of learning 

Attentional models 

• attentional models focus on CS attention-getting

• Attention to CS depends on surprisingness of US on previous trials 

  • If US was surprising, it increases attention to CS on next trial 

  • If CS was followed by expected US, don't pay attention to CS on next trial 

  • More future directed 

• Assume surprisingness of US will alter attention paid CSs on future trials

Relative-waiting-time hypothesis 

• Comparison between how long you have to wait for US during CS

  • VS

• How long you have to wait for the US during ITI (intertrial interval) (time between two CS-US pairings) 

Relative-waiting-time: example

• If wait time from CS to US is short → CS is then very useful in preparing us how to act

  • Therefore = CR is very high bc CS was so useful

• If the wait time from CS to US is long (ex. As long as ITI) → the CS does not provide enough information 

  • Therefore = CR is low 

• Too long of time in between CS-US  → more confusing than helpful 

Comparator hypothesis

• There are other cues available to the animal while in the experimental context 

• Hypothesis assumes: 

  • whether conditioned excitatory or inhibitory conditioning occurs depends on relative strengths of excitatory value of CS (target)

  •  vs.

  • comparators (context)

Comparator cues

other cues present when the target CS is being conditioned

comparator: inhibitory responding

if excitatory value of context > excitatory value of target

comparator: excitatory responding

excitatory value of context < excitatory value of target

Comparator hypothesis and blocking

• The CH states that what is blocked is responding to CS_B 

  • “We have ability to learn ab CS_B, it is just based on how power it is

• How do we test this?

  • Remove the block to CS_B 

  • Revaluation effect

Revaluation effect

• how to remove block by changing the conditioned value of CS

1. Normal blocking procedure

2. Eliminate responding to CS_A

   1. Extinction of CS_A 

3. Test for CR to CS_B

   1. If subject responds, it shows that learning was not blocked, but performance was blocked