Classical Conditioning PDF

Background of classical conditioning Pavlov’s Classical conditioning Pre-: US food) > UR (salivation) During: US (food)+NS (tone) > UR (salivation) Post-: CS (tone) > CR (salivation) Procedures APPETITIVE (POSITIVE STIMULUS) US...

Background of classical conditioning Pavlov’s Classical conditioning Pre-: US food) > UR (salivation) During: US (food)+NS (tone) > UR (salivation) Post-: CS (tone) > CR (salivation) Procedures APPETITIVE (POSITIVE STIMULUS) US UR CS CR DOG Food salivation bell Salivation Approach/ Sexually receptive QUAIL mounting, light Approach mate copulation AVERSIVE (NEGATIVE STIMULUS) US UR CS CR FLY shock Stick Escape Odor escape Conditioned Emotional shock activity Tone Freezing Response (CER) Eyeblink Air pu blink Tone blink Eye-blink conditioning Tone + air puff pairings over time (usually slow process) Eye blink = reflex Learning can be altered 1. Conditioning 2. Acquisition 3. Consolidation 4. Retrieval f  Extinction learning: non-reinforced presentations of the conditioned cue Possible causes: Spontaneous recovery Renewal Reinstatement Pathology Contiguity vs Contingency Contiguity: pairing needs to be close in time for association to be made Bower and Trabasso showed that humans don’t learn superfluous information about cues (not just paired in time-contiguity) Cues must provide predictive information about the outcome in order to be learned about Content is what assists learning Contingency: pairing needs to be close in space for association to be made Kamin’ study Phase 1: Paired CS-US until association learned Controls: were not pre-trained Phase 2: added a 2nd CS (redundant) Test Results: animals pre-trained do not learn about the 2nd cue while controls did learn that the bell was predicting the food Blocking effect: learning about a cue does not occur if it is redundant Surprising events are unexpected (not predicted) and should lead to new learning in order to better predict these events Rescorla-Wagner Model Prediction errors: Mismatches between predictions and actual experience are called the discrepancy between actual and expected outcomes Associative strength [V]: the strength of the association/predictive value of the CS to the US ΔV= ( – V) - : CS learning rate - : US learning rate - : Max conditioning capable of the US (usually 1, but can change for unblocking procedures) 𝛌 𝛃 𝛂 𝛂 𝛃 𝛌 𝚺  Overshadowing: Cues compete for associative strength [ V] (compete for salience) - calculating the associative strength for both stimuli - Sum V for both and recalculate (max = 1) Mechanisms of learning (according to RW model) Each CS develops its own associative strength with a US The expectation of the US is based on the pooled/summed associative strength of all CSs present Learning (changes in associative strength) are proportional to the size of the prediction error How RW model explains: Blocking Test group: ΔVa: starts very high ΔVb: null ΔVx: is very low, since A already predicts Controls: ΔVa: increases moderately but not as high as CSb ΔVb: starts and stays high ΔVx: learn a little bit about CSx (but will not be predictor = prediction error signal) Neural substrates of re exive classical conditioning Cerebellar Circuit in Eyeblink Conditioning Tone (US) > Pons > Cerebellum Air puff (CS) > Inferior olive Purkinje cells: integrate information about the CS and US Nictitating membrane(nm): opaque, eye covering to protect from environment Extracellular electrophysiology: record voltage signal to identify times of neuron firing fl 𝚺 Recording in Interpositus Nucleus (IN) & Purkinje cells PRE- NM not showing a response to NS (tone) Cells show increased in response to the US (puff) after presentation DURING Conditioned responses to the puff develop over several days develop strong responses to the US POST- Responses to the CS (tone) are much larger than responses to the US (puff) Still get responses to US by itself NM closes in anticipation to the puff HOW INTERPOSITUS NUCLEUS DEVELOPS RESPONSES TO THE CS Purkinje cells become inhibited by the CS inhibitory cells PRE: little response to CS/ inhibited by US Inhibit downstream neurons (Interpositus nucleus) DURING: reduction in firing to the CS (disinhibition) allow Interpositus nucleus to fire more Inferior Olive: Neurons for learning If you stimulate the neurons in the inferior olive (mimic the air puff), you get learning to the US/CS - the CR to the US/CS will be the same ***neurons in inferior olive is required for learning Climbing Fibers: Error signal Neurons in the inferior olive are inhibited by the interpositus nucleus Climbing fibers: activated by unexpected US inhibited by expectation of US Prediction error: Actual - expected Error signals from climbing fibers adjust synaptic strength on the parallel fibers and the Purkinje cells - when US is expected the interpositus nucleus from Error-related activity in the inferior olive PRE-: no response to CS/response to US DURING: no response to CS, no response to US TEST (no CS): increased response to US *** neurons only respond when US is unexpected - predicted and omitted - Not predicted and presented Damage to the circuit Lesion in interpositus nucleus Inability to express the CR (eyeblink) Lesion in Inferior Olive ** inhibitory input from IN onto the IO before conditioning: CR will never be learned Because it blocked the transmission to the Purkinje cells are not inhibited by CS After conditioning: extinction of CR even when US is presented because the prediction signal is removed Questions Is it possible to causally distinguish whether the Inferior Olive carries information about US from information related to a prediction error? Blocking Design: If a CS was already implemented to be a predictor , a 2nd US cannot be implemented as a predictor of the same outcome Compete for the same associative strength Is the function of the Inferior Olive consistent with a prediction error? Hypothesis: The inhibitory input from the interpositus to inferior olive is critical for the blocking effect. Methods: Use of GABA antagonist to block the inhibitory signal from IN to IO ACSF: neutral injection in IO PTX: GABA-antagonist into IO Controls: no training phase Tone conditioning: ACSF and PTX learning between CS and CR Compound phase: no change between ACSF and PTX group (i.e. GABA is not involved in the learning phase) Test phase: ACSF: no learning to compound US until L5 PTX: Decreased learning but similar to controls Controls: learning to US