1. Theories of Classical Conditioning

2. Critical CS-US relationship Important (critical) things to note about classical conditioning: – the CS MUST precede the US – the CS MUST predict the US – if the CS does not predict the US, no conditioning occurs – the CR does not have to be identical to the UR E.g., subtle differences even Pavlov noticed) may even be opposite: Morphine studies Any response is a classically conditioned response if it – occurs to a CS – after that CS has been paired with a US – but does NOT occur to a randomly presented CS-US pairing

3. Theories: WHY do organisms respond to predictability? Pavlov: Stimulus substitutability theory Kamin: Surprise theory Rescorla and Wagner: Computational Model

4. Pavlov: Stimulus Substitution Theory – Basic premise of theory w/repeated pairings between CS and US, CS becomes substitute for the US thus, the response initially elicited only by US is now also elicited by CS – sounds pretty good: salivary conditioning: US and CS both elicit salivation eyeblink conditioning: both elicit eyeblinks – Theory was doing well until we found compensatory CRs

5. Pavlov: Stimulus Substitution Theory – Criticisms and Flaws: CR is almost never an exact replica of the UR an eyeblink to UR of air puff = large, rapid closure eyeblink to CS of tone = smaller, more gradual closure Defense of theory: Hilgard (1936): Why differences in CR and UR: – intensity and stimulus modality of the CS and US are different – Thus: differences in Response magnitude and timing are to be expected – But still doesn’t explain OPPOSITE CR

6. Pavlov: Stimulus Substitution Theory BIGGER PROBLEM: – whereas many US's elicit several different R's, as a general rule not all of these R's are later elicited by the CS E.g. Zener (1937) – dog presented w/food as US: found that the dog elicited a number of UR responses to the food E.g., salivation, chewing, swallowing, etc. – CS not elicit all of those responses NO CRs of chewing and swallowing Just the CR of just salivation on other hand: CR may contain some of responses that are not part of CR: – Zener found that dogs turned head to bell – But no head turns to presentation of food

7. Modifications of SST MODIFICATIONS OF SST: (Hilgard) – only some components of UR transferred to CR – CS such as a bell often elicits unconditioned responses of its own, and these may become part of CR SIGN TRACKING: Hearst and Jenkins 1974 – emphasized this change in form of CR vs. UR – Also Jenkins, Barrara, Ireland and Woodside (1976) Sign Tracking : animals tend to – orient themselves toward – approach – explore any stimuli that are good predictors of important events such as the delivery of food

8. Set up: 1.Initial training: Light turns on above feeder  feeder releases pieces of hot dog 2.Test: a.Light turns on above feeder, then above each of the other walls b.Forms a sequence of 1  2  3  4 3. What is optimal response? 4. But: Dog “tracked the sign” 1 2 3 4 Jenkins, Barrara, Ireland and Woodside (1976)

9. Modifications of SST Strongest data against SST theory: Paradoxical conditioning – CR in opposite direction of UR Black (1965): – heart rate decreases to CS paired w/shock – US of shock elicits UR of heart rate INCREASE – But CS of light or tone elicits CR of heart rate DECREASE Seigel (1979): conditioned compensatory responses – Morphine studies – evidence of down regulation in addiction – Actual cellular process in neurons (and other cells, too!) – thus SST theory appears incorrect

10. Perceptual Gating Theory Perceptual gating theory: – Idea that only if CS is biologically relevant will it get processed – If a CS doesn’t get processed it can be predictive/informative – Animals attend to biologically relevant stimuli Problem: – Data show that under certain circumstances a stimulus is “attended to” or “processed”, but still does not serve as a CS with an accompanying CR – Issue remains: is the stimulus the most predictive? – Second issue: Defining “biologically relevant”

11. Kamin’s work: 1967-1974 Blocking and overshadowing Overshadowing: – use one "weak" and one "strong" CS – CS1+CS2  US – reaction to weaker stimulus is blotted out by stronger CS – Demonstrated by Pavlov Blocking: – Train 1 CS, then add a second CS to it: CS1  US CS1+CS2  US – test each individually after training – Find that only one supports a CR – One stimulus “blocks” learning to second CS – Demonstrated by Kamin

12. Kamin’s blocking experiment used multiple CS's and 4 groups of rats the blocking group receives – series of L+ trials which produce strong CR – series of L+T+ trials – then tested to just the T control groups receives – SAME TOTAL NUMBER OF TRIALS AS BLOCKING GROUP – no first phase – L+ only; Test T – T+ only; Test T – LT+ only: Test T

13. Kamin’s blocking experiment prediction: since both received same # of trials to the tone- should get equal conditioning to the tone results quite different: Blocking group shows no CR to the tone- the prior conditioning to the light "blocked" any more conditioning to the tone directly contradicts frequency principle (remember associationism!) Group Phase I Phase II Test Phase Result Control --- L+ T T elicits no CR Control --- T+ T T elicits CR Control --- LT+ T T elicits a CR Blocking L+ LT+ T T elicits no CR

14. Things we know about blocking: the animal does "detect" the stimulus: – can’t be perceptual gating issue – EXT of CR with either T alone or with LT – EXT occurred faster with compound LT appears to be independent of: – length of presentation of the CS – number of trials of conditioning to compound CS constancy of US from phase 1 to 2 important!!!! – US must remain identical between the two phases or no blocking influenced by: – Type of CR measure (used CER, not as stable as non fear CR) – nature of CS may be important- e.g. modality – intensity of CS or US stimuli important depends on amount of conditioning to blocking stimulus which already occurred

15. Change in either US or CS can prevent/ overcome blocking Change the intensity of the CS from phase 1 to phase 2 – Overshadowing could be playing a role – strong vs weak stimulus – e.g. experiments when changed from 1 ma to 4 ma shock – quickly condition to compound stimulus – little or no overshadowing or blocking Change in intensity of either CS stimulus- – Change in context from Phase 1 to Phase 2 lT then T Lt then T – presents a different learning situation and no blocking Any ideas about what is happening?

16. Explanations of Blocking: Poor Explanation: Perceptual gating theory: – tone never gets processed – tone not informative – data not really support this (evidence that do “hear” tone) Good Explanation: Kamin's Surprise theory: – to condition requires some mental work on part of animal – animal only does mental work when surprised – bio genetic advantage: prevents having to carry around excess mental baggage – thus only learn with "surprise" – situation must be different from original learning situation Better Explanation: Rescorla Wagner model: – particular US only supports a certain amount of conditioning – if one CS “hogs” all that conditioning- none is left over for another CS to be added – question- how do we show this?

17. Recorla: Which is more important? CS-US correlation vs. contiguity CS-US contiguity: – CS and US are next to one another in time/space – In most cases, CS and US are continguous CS-US correlation: CS followed by the US in a predictive correlation: If perfect correlation (most predictive)- most conditioning p(US/CS) = 1.0 p(US/no CS) = 0.0 But: life not always a perfect correlation

18. CS-US correlation is more critical Rescorla (1966, 1968): Showed how 2 probabilities interact to determine size of the CS – CS = 2 min tone; presented at random intervals (M = 8 min) – for: Group 1: p(shock/CS) = 0.4 during 2 min presentation – For Group 2: p(shock/no CS) = 0.2 Which group should show more conditioning? WHY?

19. Robert Rescorla (1966) Examined predictability 6 types of Groups CS-alone – present CS alone with no US pairing – problem: not have same number of US trials as experimental animals do, may actually be extinction effect Novel CS group: – looks at whether stimulus is truly "neutral" – may produce habituation- animal doesn't respond because it "gets used to it" US-alone – present US alone with no CS pairing – problem: not have same number of CS trials

20. Rescorla: 6 types of control groups Explicitly unpaired control – CS NEVER predicts US – that is- presence of CS is really CS-, predicts NO US – animal learns new rule: if CS, then no US Backward conditioning: – US precedes CS – assumes temporal order is important (but not able to explain why) – again, animal learns that CS predicts no US Discrimination conditioning (CS+ vs CS-) – use one CS as a plus; one CS as a minus – same problem as explicitly unpaired and backward- works, but can work in certain circumstances (taste avoidance)

21. Rescorla: Results with 6 Groups CS-alone: No conditioning, but habituation to CS Novel CS group: novel worked better than CS with previous experience. US-alone: habituation to CS Explicitly unpaired control: – Got GREAT conditioning – Learned that the CS NEVER predicts the US! Backward conditioning: – US preceded CS – assumed temporal order is important – It was: Animal learned that CS predicts NO US, but US predicted CS Discrimination conditioning (CS+ vs CS-) – use one CS as a plus; one CS as a minus – Got discrimination – Animals paid attention to whatever stimulus was MOST PREDICTIVE!

22. CS-US correlation: Summary of Results whenever p(US/CS) > p(US/NO cs): – CS = EXCITATORY CS – that is, CS predicts US – amount of learning depended on size difference between p(US/CS) and p(US/no CS) whenever p(US/CS) <p(US/NO CS): – CS = INHIBITORY CS – CS predicts ABSENCE of US – amount of learning depended on size difference between p(US/CS) and p(US/no CS) whenever p(US/CS) = p(US/NO cs): – CS = NEUTRAL CS – CS doesn’t predict or not predict CS – no learning will occur because there is no predictability.

23. CS-US correlation vs. contiguity Thus: appears to be the CORRELATION between the CS and US, not the contiguity (closeness in time) that is important Can write this more succinctly: – correlation carries more information – if r = + then excitatory CS – if r = - then inhibitory CS – if r = 0 then neutral CS (not really even a CS)

24. Classical condition is “cognitive” (oh the horror of that statement, I am in pain) PREDICTABILITY is critical Learning occurs slowly, trial by trial – Each time the CS predicts the US, the strength of the correlation is increased – The resulting learning curve is monotonically increasing: Initial steep curve Levels off as reaches asymptote – There is an asymptote to conditioning to the CS: Maximum amount of learning that can occur Maximum amount of responding that can occur to CS in anticipation of the upcoming US We can explain this through an equation!

25. Answers to Blocking and Overshadowing Overshadowing: – use one "weak" and one "strong" CS – reaction to weaker stimulus: less CR – Reaction to stronger stronger stimulus: more CR Blocking: – What is being predicted – Does LT give any more information/predictability than L alone? – If not, then L “blocks” learning to LT

26. Assumptions of Rescorla-Wagner (1974) model Model developed to accurately predict and map learning as it occurs trial by trial Assumes a bunch of givens: – Assume animal can perceive CS and US, and can exhibit UR and CR – Helpful for the animal to know 2 things about conditioning: what TYPE of event is coming the SIZE of the upcoming event Thus, classical conditioning is really learning about: – signals (CS's) which are PREDICTORS for – important events (US's)

27. Assumptions of R-W model assumes that with each CS-US pairing 1 of 3 things can happen: – the CS might become more INHIBITORY – the CS might become more EXCITATORY – there is no change in the CS how do these 3 rules work? – if US is larger than expected: CS = excitatory – if US is smaller than expected: CS= inhibitory – if US = expectations: No change in CS The effect of reinforcers or nonreinforcers on the change of associative strength depends upon: – the existing associative strength of THAT CS – AND on the associative strength of other stimuli concurrently present

28. More assumptions Explanation of how an animal anticipates what type of CS is coming: – direct link is assumed between "CS center" and "US center": e.g. between a tone center and food center In 1970’s: other researchers thought R and W were crazy with this idea Now: neuroscience shows formation of neural circuits! – assumes that STRENGTH of an event is given the conditioning situation is predicted by the strength of the learned connection – THUS: when learning is complete: the strength of the association relates directly to the size or intensity of the CS Asymptote of learning = max learning that can occur to that size or intensity of a CS Maximum amount of learning that a given CS can support

29. More assumptions The change in associative strength of a CS as the result of any given trial can be predicted from the composite strength resulting from all stimuli presented on that trial: – Composite strength = summation of conditioning that occurs to all stimuli present during a conditioning trial – if composite strength is LOW: the ability of reinforcer to produce increments in the strength of component stimuli is HIGH More can be learned for this trial – if the composite strength is HIGH: reinforcement is relatively less effective (LOW) Less can be learned for this trial- approaching max of learning

30. More assumptions: Can expand to extinction, or nonreinforced trials: – if composite associative strength of a stimulus compound is high, then the degree to which a nonreinforced presentation will produce a decrease in associative strength of the components is LARGE – if composite associative strength is low- nonreinforcement effects reduced

31. The Equation!: Yields an equation: THE Rescorla Wagner (1974) model!!!!! V i =α i ß j (Λ j -V sum ) V i = amount learned (conditioned) on a given trial Α i = the salience of the CS ß j = the salience of the US (Λ j -V sum ) = total amount of conditioning that can occur to a particular CS-US pairing What does this equation say? The amount of conditioning that will occur on a given trial is a function of: The size of the salience of the CS multiplied by The size of the salience of the US multiplied by (The maximum amount of learning minus the amount of learning that has already occurred).

32. Let’s use this in an example: First example: A rat is subjected to conditioned suppression procedure: – CS (light) ---> US (1 mA shock) – Question: what is associative strength? – 1 = associative strength that a 1mA shock can support at asymptote ( λ j ) (I am arbitrarily setting this value for easy math) So, we will say that the associative strength of a 1 mA shock = 100 units of association/learning – V L = associative strength of the light (strength of the CS-US association) thus: λ 1 = animal’s maximum reaction to the size of the observed event (actual shock) V L = measure of the Subjects current "expectation" about the light predicting the light. V L will approach λ 1 over course of conditioning: V L = λ 1

33. First trial: CS L  US shock CS (light+tone) --> 1 mA shock on trial 1 (no previous pairing) – Λ j = max amount of conditioning that can occur to the CS L : Let’s set it at 100 – V sum = assoc. strength of all paired trials so far (0) – Can set α i = 0.5 – Can set ß j = 1.0 – V L = α i ß j (Λ j -V sum ) just plug in numbers – V L = 0.5*1.0(100-0) = 50 units of conditioning/learning

34. Second example: 2CS's: CS (light+tone) --> 1 mA shock – V sum = V L + V T = assoc. strength of the 2 CS's – (still 0 on trial 1) – V sum = α i ß j (λ n ) – if V L and V T equally salient: V L = 0.5α i ß j; V T = 0.5α i ß j – V T = 0.5*0.5*(100-0) = 25 units of learning

35. WHY is this equation important? We can use the three rules to make predictions about amount and direction of classical conditioning λ j > V sum = excitatory conditioning – The degree to which the CS predicted the size of the US was GREATER than expected, so you react MORE to the CS next trial λ j < V sum = inhibitory conditioning – The degree to which the CS predicted the size of the US was LESS than expected, so you react LESS to the CS next trial λ j = V sum = no change: – The CS predicted the size of the US exactly as you expected

36. Now have the Rescorla-Wagner Model: Model makes predictions on a trial by trial basis for each trial: predicts increase or decrement in associative strength for every CS present Can specify amount and direction of the change in conditioning!

37. Now have the Rescorla-Wagner Model: Restate the equation: V i =α i ß j (λ j -V sum) V i = change in associative strength that occurs for any CS, i, on a single trial λ j = associative strength that some US, j, can support at asymptote V sum = associative strength of the sum of the CS's (strength of CS-US pairing) α i = measure of salience of the CS (must have value between 0 and 1) ß j = learning rate parameters associated with the US (assumes that different beta values may depend upon the particular US employed)

38. Can say this easier! How much you will learn on a given trial (Vi) is a function of: – α i or how good a stimulus the CS is (how well it grabs your attention) – ß j or how good a stimulus the US is (how well it grabs your attention – Λj or how much can learning can be learned about the CS-US relationship – AND V sum or how much you have learned ALREADY!

39. Okay, you got all that? Let’s put this baby to work…….. …….we will try a few examples

40. The equation: Vi =αißj(λ j-Vsum) V i = change in associative strength that occurs for any CS, i, on a single trial α i = stimulus salience (assumes that different stimuli may acquire associative strength at different rates, despite equal reinforcement) ß j = learning rate parameters associated with the US (assumes that different beta values may depend upon the particular US employed) V sum = associative strength of the sum of the CS's (strength of CS-US pairing) λ j = associative strength that some CS, i, can support at asymptote In English: How much you learn on a given trial is a function of the value of the stimulus x value of the reinforcer x (the absolute amount you can learn minus the amount you have already learned).

41. Acquisition first conditioning trial: Assume (our givens) – CS = light; US= 1 ma Shock – V sum = V l; no trials so V l = 0 – thus: λ j -V sum = 100-0 = 100 – -first trial must be EXCITATORY BUT: must consider the salience of the light: – α i = 1.0 – ß j = 0.5

42. Acquisition first conditioning trial: CS = light; US= 1 ma Shock – V sum = V l; no trials so V l = 0 – thus: λ j -V sum = 100-0 = 100 – -first trial must be EXCITATORY BUT: must consider the salience of the light: α i = 1.0 and learning rate: ß j = 0.5 Plug into the equation: for TRIAL 1 – V L = (1.0)(0.)(100-0) = 0.5(100) = 50 – thus: V only equals 50% of the discrepancy between A j an V sum for the first trial

43. Acquisition Plug into the equation: – for TRIAL 1 – V L = (1.0)(0.)(100-0) = 0.5(100) = 50 – thus: V L only approaches 50% of the discrepancy between A j and V sum is learned for the first trial

44. Acquisition TRIAL 2: – Same assumptions! – V L = (1.0)(0.5)(100-50) = 0.5(50) = 25 – V sum = (50+25) = 75

45. Acquisition TRIAL 3: – V L = (1.0)(0.5)(100-75) = 0.5(25) = 12.5 – V sum = (50+25+12.5) = 87.5

46. Acquisition TRIAL 4: – V L = (1.0)(0.5)(100-87.5) = 0.5(12.5) = 6.25 – V sum = (50+25+12.5+6.25) = 93.75 TRIAL 10: V sum = 99.81, etc., until reach ~100 on approx. trial 14 When will you reach asymptote?

48. How to explain overshadowing? Yep, it is good old Rescorla-Wagner to the rescue!

49. Remember Overshadowing Pavlov: compound CS with 1 intense CS, 1 weak – after a number of trials found: strong CS elicits strong CR – Weak CS elicits weak or no CR Note: BOTH CSs are presented at same time – Why would one over shadow or overpower the other? – Why did animal not attend equally to both?

50. Overshadowing Rescorla-Wagner model helps to explain why: Assume – α L = light = 0.2; α T = tone = 0.5 – ß L = light = 1.0 ; ß t = tone = 1.0 Plug into equation: – V sum = V l + V t = 0 on trial 1 – V L = 0.2(1)(100-0) = 20 – V t = 0.5(1)(100-0) = 50 – after trial 1: V sum = 70

51. Overshadowing TRIAL 2: – V L = 0.2(1)(100-(50+20)) = 6 – V t = 0.5(1)(100-(50+20)) = 15 – V sum = (70+(6+15)) = 91 TRIAL 3: – VL = 0.2(1)(100-(91)) = 1.8 – Vt = 0.5(1)(100-(91)) = 4.5 – Vsum = (91+(1.8+4.5)) = 97.3 and so on – thus: reaches asymptote (by trial 6) MUCH faster w/2 CS's NOTE: CS t takes up over 70 units of assoc. strength CS l takes up only 30 units of assoc. strength

52. Overshadowing

53. Blocking Similar explanation to overshadowing: – Does not matter whether V L has more or less saliency than V t, – CS has basically absorbed all the associative strength that the CS can support Why?

54. Blocking give trials of A-alone to asymptote: – reach asymptote: V L = λ j =100 = Vsum NOW add trials to compound stimuli: – CS of the light has salience: α L =.5465 – CS of tone has salience of: ß t =0.464 – Note that CS tone has higher salience! – Eh, oh, the math is going to be TOO HARD to do!!!!!

55. Blocking Or IS the math to hard to do? First compound V 1 Trial: V t = αß(Λ j -V sum ) What is V sum after the training to the CS light? That’s right V sum = ___________ V t =0.*1.0*(100-100)= _____________ No learning!

56. How could one eliminate blocking effect? increase the intensity of the US to 2 mA with λ j now equals = 160 – Learning so far: V sum still equals 100 (learned to 1 mA shock) – But now: TOTAL learning is increased to 160!

57. How could one eliminate blocking effect? plug into the equation: (assume V l and V t equally salient) – V t = 0.2(1)(160-100) = 0.2(60) = 12 – V l = 0.2(1)(160-100) = 0.2(60) = 12 – V sum = 100+12+12 =124

58. How could one eliminate blocking effect? on trial 2: – V sum = 124 – V t = 0.2(1)(160-124) = 0.2(36) = 7.2 – V l = 0.2(1)(160-124) = 0.2(36) = 7.2 – V sum now = (124+14.4) = 138. – Again, monotonically increasing curve. Thus, altering the salience of the US alters the learning Does altering the CS make the same change?

59. Can also explain why probability of reward given CS vs no CS makes a difference: π = probability of US given the CS or No US given No CS can make up three rules: – if π ax > π a then V x should be POSITIVE – if π ax < π a then V x should be NEGATIVE – if π ax = π a then V x should be ZERO modified formula: (assume λ 1 =1.0; λ 2 =0; ß 1 =.10; ß 2 =.05; α 1 =.10; α 2 =.5) Πa = probability of reward.

60. Critique of the Rescorla-Wagner Model: R-W model really a theory about the US effectiveness: – says nothing about CS effectiveness How WELL a CS predicts as a combo of salience and probability – states that an unpredicted US is effective in promoting learning, whereas a well-predicted US is ineffective Reason has to do with brain processing of all of this

61. Critique of the Rescorla-Wagner Model: Fails to predict the CS-pre-exposure effect: – two groups of subjects (probably rats) – Grp ICS-US pairingsControl – Grp II CS aloneCS-US pairings PRE-Expos Bob and Tom effect – Bob always hangs with Tom – You are dating Tom – You have a BAAAAAD breakup with Tom – Now you hate Bob….why?

62. Critique of the Rescorla-Wagner Model: In pre-exposure effect, simply being around a neutral stimulus alters its ability to become conditioned Original R-W model doesn't predict any difference, – Assumes no conditioning trials occur when CSs presented in absence of US so V sum = 0 – This appears to be wrong Conditioning likely occurring any time 2 stimuli are together – Form an incidental association – Need to modify the equation to account for this – They have, but we won’t!

63. Critique of the Rescorla-Wagner Model: Original R-W model implies that salience is fixed for any given CS – R-W assume CS salience doesn't change w/experience – these data strongly suggest CS salience DOES change w/experience Newer data supports changes salience – data suggest that Salience to a CS DECREASES when CS is repeatedly presented without consequence – CS that is accidentally paired with another CS INCREASES in salience – NOW: appears that CS and US effectiveness are both highly important Model has stood test of time, now widely used in neuroscience

64. Dopamine and Rescorla Wagner Model Turns out that changes in dopamine (DA) levels in dorsal striatal limbic cortical pathway vary as we learn And guess what: these levels can be predicted by the RW model! But, once a CS-US pairing (or an operant R-S R pairing) become well learned, the circuit begins to involve lower parts of the brain – Circuit begins to involve basal striatal areas – Becomes an “automated” or mastered behavior – No longer involves being “surprised” – Only returns to this pathway if the CS-US relation change!!!!!