Transcript Lecture 4
Animal Cognition Clive D. L. Wynne Lecture 4 Cause & Effect Reasoning Cause and Effect Learning Pavlovian conditioning Aka, Classical, Respondent or Type II conditioning Ivan P. Pavlov 1849-1936 Already famous as physiologist studying digestion • Noticed dogs would salivate to sight of scientist – ‘Psychic secretions’ Pavlov’s dogs Originally Put food in dog’s mouth Dog salivates After doing this a few times Dog salivates just when it senses that food is going to be put in mouth Classical conditioning A process by which an initially neutral stimulus (the conditioned stimulus, CS) comes to produce a response (the conditioned response, CR) through being repeatedly paired with a stimulus (the unconditioned stimulus, US) that produces a response prior to training (the unconditioned response, UR). Classical conditioning Food Salivation US UR Metronome No salivation CS - Conditioning… Metronome: Food Salivation CS UR After conditioning… Metronome CS Salivation UR Pavlovian terminology Unconditioned Stimulus (US) – stimulus that produces a response prior to training Unconditioned response (UR) – response to the US, occurs prior to training Conditioned stimulus (CS) – initially neutral stimulus Conditioned response (CR) – response to the CS that only develops after CS and US have been repeatedly paired. Pavlovian procedures Before conditioning Conditioning US-food in mouth UCR salivation US-food in mouth CS-tone Orientation but no salivation CS-tone UCR salivation After conditioning US-food in mouth CS-tone UCR salivation CR salivation Conditioned Flavor Aversions Eat something Feel sick Avoid that flavor in future CFA Experiment Group Stimulus Outcome Group A Bright-noisy water Foot shock Group B Bright-noisy water Sickness OK Group C Tasty water Foot shock OK Tasty water Sickness Group D Test water Avoid Avoid CFA Experiment Some consequences can be more readily associated with certain stimuli Interval between CS and US can be very long Common experience is a form of learning Functions of CRs Prefiguring CRs enable individual to optimize interaction with upcoming US. Evidence in favor: Dog saliva has different composition CR saliva is thinner than UR – prepares digestive system Rats jump to shock US: freeze to CS Jump if snake attacks; freeze if snake detected Functions of CRs Predator defense CSs that predict danger USs evoke predator defense reactions Rats turn and flee from danger CS But will freeze if learn that there is no place to run to Autonomic responses: increase heart rate etc. Diversity of Pavlovian cond. Animals learn about what predicts what Paramecia lean to withdraw when vibrations are followed by electric shock Honeybees associate odors, colors & shapes with sucrose; but not with escape from box. Marine snails (aplysia) associate light mantle touch with shock to tail (close gill flap) Lemon sharks learn to blink to light that predicts mild shock close to eye – likewise goldfish. Diversity of Pavlovian cond. Siamese fighting fish attack own reflection. Conditioned to attack red light. Leopard frogs conditioned to blink to light tough on nostril followed by touch on eye – also toads Bengal monitor lizards conditioned with flickering light CS to food US attacked the light. Collared lizards changed breathing, pulse and leg movements in response to sound and light that preceded electric shock. Birds: especially pigeons Mammals: especially rats, mice, rabbits, dogs & humans; also cats and marsupials. Diversity of Pavlovian cond. Hollis et al., 1997. Male blue gouramis attack intruder males and may repel egg-bearing females Males exposed to 10-s white light followed by 5min exposure to female fish. Came to make mating displays to light – were better prepared for female – less aggressive. At end of experiment Experimental group males had sired > 1000 offspring each. Control group < 50 offspring each. Behavioral Consequences In Pavlovian Conditioning behavior is elicited by stimuli (CR and UR) But often behavior is controlled by its consequences Operant behavior The behavior operates on the environment A.k.a. instrumental conditioning The behavior is instrumental in achieving some consequence. Edward Thorndike (18741949) Put animals in boxes and watched how long it took them to escape Essentially random behavior Learning by trial and error Thorndike’s Puzzle Boxes Thorndike’s law of effect Consequences of an act determine whether it will be repeated in the future Behaviors followed by positive consequences are more likely to be repeated – strengthened – stamped in Behaviors followed by negative consequences are less likely to be repeated – weakened – stamped out Small’s Hampton Court Maze Modern mazes Skinner’s Learning by Consequences Free Operant technique Subject free to make response at any time – not driven by reflex or structure of apparatus (e.g., maze). Operant conditioning Type of learning where future probability of a behavior is affected by its consequences Skinner’s Learning by Consequences Diversity of Operant cond. More difficult to operant condition than Pavlovian because of need to find suitable reinforcement v. difficult to operant cond. cold-blooded animals. Tadpoles learnt to move away from a light to avoid a shock Anoles (lizards) learnt to escape shock by running to another part of apparatus – also collard lizards and desert iguanas Queen triggerfish pressed plastic rod to drop food into tank – also goldfish & koi carp Birds: pigeons; doves; chickens; gulls & quail Mammals: rats, mice, rabbits, cats, dogs, raccoons, skunks, ferrets, minks, farm animals, marsupials, primates, humans. Reasoning Spatial Reasoning Tolman & Honzig, 1930 3 paths – blockable at points shown. Once familiar with the maze, rats always chose the shortest remaining path after a blockage Spatial Reasoning Wolfgang Köhler, 1925 Transparent barriers (fence of wire netting) Dog, girl just > 1 yr, chickens, chimpanzees goal start Spatial Reasoning Bruno Poucet et al., 1983 Cats choose shorter path But only if barrier opaque. In extended tests cats only take shorter route if does not involve larger original detour angle. Reasoning about Gravity Hood et al., 1999 Insight Wolfgang Köhler, 1921 Chimpanzees on Tenerife in WW1 Banana hung out of reach on roof Köhler impressed by the flash of insight All six apes vainly endeavored to reach the fruit by leaping up from the ground. Sultan soon relinquished this attempt, paced restlessly up and down, suddenly stood still in front of the box, seized it, tipped it hastily straight towards the objective, but began to climb upon it at a (horizontal) distance of ½ meter, and springing upwards with all his force, tore down the banana. More Insight Pigeon Insight Robert Epstein et al., 1984 Pigeon reinforced for moving box to a green spot in the center of the chamber On other trials for standing on a box already placed in the center of the chamber and pecking at the plastic banana hanging from the ceiling. In test session pigeon was faced with the problem of pecking the banana while the box was at the side of the chamber and there was no green spot to move the box to. Pigeon quickly solved the problem by moving the box to the center of the chamber and mounting it to peck the banana. Insight Elisabetta Visalberghi & Luca Limongelli, 1994 Only 1 of 4 Capuchins learnt to avoid the trap over 140 trials. When trap was inverted still treated it as dangerous. 2 of 5 chimps successful Only 1 chimp behaved appropriately when the trap was moved up and down the tube. Tool use Ant-lions throw sand on prey that comes in their pits Japanese macaques use water to separate sand from food grains Open mollusk shells Sea otter smash them on stones held against their chests Gulls drop them onto stones from a height Hermit crabs grab sea anemones and use them to repel enemies Tool use/creation Tetsuro Matsuzawa. Bossou chimpanzees hammer/anvil nut cracking 1 on map Other chimp communities have different tools Termite fishing Water Tool creation Series Reasoning: Transitive Inferences Bryant & Trabasso (1974) Children A B C D E Kate is cleverer than May: May is cleverer than Jane. Who is the cleverest--Jane, Kate or May ? Squirrel Monkeys REWARDED NON-REWARDED + - + - + - + - McGonigle & Chalmers (1977) Pigeon Stimuli + - Pigeon Data Percent correct responses 100.00 Phase 1 Phase 2 Phase 3 Phase 4 90.00 80.00 70.00 60.00 Pigeons Model 50.00 A + B - B + C - C + D - D + E - A + B - B + C - C + D - D + E - X A B C + + + + A B C D - - - Pair/ Phase D + E - E + F - X + A - A + B - B + C - C + D - D + E - E + F - F + X - Counting: Sheba (chimp)