Behaviour #5: learning

• changes in behavior that results from an animal's interaction with its environment
• More precise:
• -specific to the event experienced
• -adaptive: species that adapt to learn will survive more than others
• -lasting
• -change of, modification in, or creation of, a single behavior
• -involving the nervous system
• -as a result of experience with an external event or series of events in an individuals life
• -excludes fatigue, hormonal changes, other developmental or maturational processes, injury

• no you can only see the results
• (see expression of responses as a result of learning)

the changes in the nervous system by which information is stored during learning

• coincidence: 2 events happen together by chance
• contingency: two events happen together more often than they should by chance

predictable departures from randomness

• Brains are like contingency detecting and exploiting machines
• -perception is the filtering of info from randomness
• -learning is the ability to adapt responses to new contingencies (info) that are not programmed in the genes

• Ethologists
• Comparative psychologists

• function that learning plays in the animal's natural environment
• phylogenetic differences in learning abilities
• role of leaning in development

• elucidate mechanisms underlying learning
• general "law" of leaning

• Non-associative learning: animal is exposed repeatedly to a single stimulus, and behavior changes
• Associative learning: animal is exposed to 2 or more stimuli that have a particular relationship to one another. Animal demonstrate associative learning if response to one stimulus is altered by exposure to the second.
• complex learning: insight, reasoning

• waning of a response, when a stimulus is repeated, even though the full response could still be made
• most primitive and universal form of learning
• eg a dog in a new home may be startled by the sound of a door opening, but over time this startle response slowly disappers

• Aplysia (marine slug)
• -given a tactile stimulus
• weak stimulus: took longer to habituate to than a strong stimulus
• (however maybe the muscles in the slow were just too exhausted to respond?)
• after a rest full recovery of response is seen

recover of response strength due to a strong novel stimulus (i.e. increase intensity)

• Association between one stimulus, and either another stimulus or the animals own behavior, leads to a change in behavior
• ex. pigs in a swine unit may associate feed with the sound of the feed truck, time of day, stockman's presence, and begin salivating before food actually arrives
• ex. rats in a maze may associate food with the act of entering a particular corridor, and enter that corridor repeatedly each time they run this maze
• ex cows become nervous when vet truck arrives

• Classical condition:
•   -associations between events over which the animal has no direct control (ie feeding truck and food)
•   -requires multiple exposures
• Instrumental Conditioning
•   -the animal's behavior is instrumental in learning (eg choosing the correct corridor in a maze)
•   -requires multiple exposures
• Passive avoidance learning (taste aversion learning)
•   -eg rats can eat a novel food, become sick several hours later, and will avoid the food in the future
•   -probably very different from classical and instrumental conditioning, not least because ir requires ONE exposure

Blue Jays eat monarch butterfly's and then throws up (so don't eat again)

• unconditional stimulus (UCS): puff of air
• unconditional response (UCR): eye blink
• neutral stimulus (NS): tone "sound"UCS and NS are presented together after repeated presentations, the animal responds to both UCS and NSNS is NOW referred to as a conditional stimulus (CS) and UCR is termed the conditional response (CR)

• an Existing behavior is elicited by a new stimulus neutral stimulus
• can result in unconditioned response if paired with UCS
• long lasting
• "pavlovian learning"

• the CS can act like an UCS for a new NS
• ie a flashing light (NS) can be paired with a bell (CS) and the dog will salivate to the flashing light

Disrupt association between UCS and CS-present CS with no UCSRing bell without giving dog food -so prevent a stimuli without supporting it

• Classical Conditioning (pavlovian learning)
• -the animal has no control over events that are changing its environment
• -it just comes to respond to one event as if it were another
• Instrumental conditioning (instrumental learning, operant conditioning, trial and error learning
• -animal can behave spontaneously, its behavior is instrumental in the learning process
• -learn from results of action

• This is occurring all the time, even when you don't want it to!
• -dog jumps up and receives social interaction
• -cat scratching furniture

• learns more readily if response occurs spontaneously rather than forced (pushing on back end)
• food reward in hand, lift hand up, with one hand behind puppy to prevent 'backing up' and puppy naturally sits
• as puppy sits say sit and reward (reinforce) immediately (food and praise are the reinforcing stimuli - DO NOT reward non sitting
• eventually raised hand becomes visual clue for sitting; just as "SIT" becomes an auditory one

animal is active and specifically involved in process

• believed that instrumental conditioning was behind most if not all complex behaviour: his point was pivotal to modern behaviorism (because almost all complex acquired behavior could in theory be explained using instrumental learning)
• Skinner referred to instrumental conditioning as operant conditioning
• -skinner box: very simple barren enviro to study operant conditioning in pure, uncorrupted form
• ie mouse has a lever which puts food in

• Operant- behavior that the experimenter wishes to alter the chance of being performed
• (operants ie pecking key, pressing lever, sitting)
• The animal must learn association between stimuli, responses, and consequences 

• shape behavior: when animal gets close to doing what should, give rewards, only reward again if gets even closer
• with shaping, very complex sequences can be taught

• works with operant behavior
• clicker associated with positive reinforce (usually food)
• Click (CS) and treat (UCS) - the click becomes classically conditioned as a secondary reinforcer
• the click is as if it were a treat
• used to shape behavior- give a click to give the animal small rewards for each approximation
• so sound became positive on its own without food

• Does shaping really make sense? How can animals learn in the wild if there is no-one there to shape them?
• in fact few operants in a skinner box reflect anything an animal would do in the wild

• For example:
• pigeons can be taught extremely complex tasks (such as telling Picasso from monet) if the operant is to peck a key, and the reward is food
• in fact they were even able to generalize from Picasso to other cubist painters and from monet to other impressionists
• BUT it was almost impossible to teach a pigeon to peck a key for a non-food reward such as access to a mate

• 1.  want behavior to increase with positive reinforcement
• 2. remove already applied stimulus to reinforce behavior
• 3. stimulus presented and the behavior decreased "ie positive punishment"
• 4. stimulus already present gets taken away and behavior decreases

• 'punishment' is used in common speech to mean "positive punishment"- performing behavior results in noxious stimulus, intended to prevent a behavior
• timing is critical
• Intensity: optimal level is the minimum required to suppress response which is very difficult to judge in practice
• positive punishments have side effects: unpredictable behavior, fear, suppression of all behaviors, makes learning harder if not right the first time

hard to apply appropriately because animal needs to associate behavior with punishment

• Animals work harder when reinforcements are
• a) few and far between
• b) relatively irregular and unpredictable

extinction- friend and foe

• continuous
• fixed interval
• variable interval
• fixed ratio
• variable ratio

**Go into notes review chart

• continuous reinforcement (CRF)
• Fixed interval (F1)
• Variable Interval (V1)
• Fixed ratio (FR)
• Variable ratio (VR)

• NO: they are simply association between stimulus and response
• ability to learn per stimulus and response is not enough to infer any sort of comprehension
• goal directed trial and error

• thinking/prediction/mental simulation
• latent learning
• cognitive maps
• learning-set learning
• insight learning
• tool use, and tool making

• Mental simulation: is the ability to run a simulation of trial-and-error learning, and to learn from it
• -obvious advantages: less costly, quicker
• -Basis of "thinking"

• Latent Learning: Putting together 2 separate experiences that have never occurred together as if they had been instrumentally conditioned
• (so taking an event that happened in the past and using that info later on, but in diff enviro)
• -may involve mental simulation
• -again obvious biological advantages

latent learning ex: ie mouse in a maze gets shocked in a black square... later on in a new maze there is a black square and it avoids it