This site is 100% ad supported. Please add an exception to adblock for this site.

PY 1 Exam Two: Learning

Terms

undefined, object
copy deck
Learning
a relatively permanent change in behavior/behavioral potential acquired through experience. A new behavior resulting from experience.
Habituation
an animal ceases to respond to a familiar stimulus.
Pavlov's three steps
1849-1936): pre-training (meat powder leads to salivation), training (meat and tone presented together), post-training (tone leads to salivation).
Historical views- Descartes
(1596-1650) believed in two kinds of behavior (Cartesian
Dualism): voluntary (controlled by the soul and consistent with free will) and involuntary (“hard-wired,” reflexes). He stated that most of animal behavior is involuntary but most of human behavior is voluntary. So, only a small percentage of human behavior could be studied objectively. Descartes believed that most of our basic mental content is inherited (Doctrine of Innate Ideas). Mind could only be studied by introspection.
British Empiricists
John Locke- “blank slate.” The empty mind comes to be filled only with contact with the outside world. So how do separate sensory experiences get organized together? Doctrine of Associationism- any sensations that occur together will become associated. (Association is a relationship between an event and a behavior.) Thus, when one sensation occurs again, it will evoke the other. Contiguity (occurring together in space and time) and frequency (associations are strengthened by repetitions) determine associations.
Contiguity
(occurring together in space and time)
Association
a relationship between an event and a behavior.) Thus, when one sensation occurs again, it will evoke the other.
conditional reflex
Pavlov showed that a reflex could be established to a new stimulus (a conditional reflex that is new from experience and is thus a kind of learning). Showed that learning could be investigated objectively, just like normal reflexes. Thus, Cartesian Dualism distinction is not necessary.
Classical conditioning
UCS (food, electric shock). UCR (salivation, limb movement. Inborn.). CS (light, tone). CR (salivation, limb movement. Acquired through learning.). UCR is stronger than the CR. Before training: UCS – UCR. Training: CS – UCS. Post training: CS – CR. Think Pavlov, lemon. Purpose of classical conditioning is to prepare body for what is about to happen. Creates new connections between stimuli. UCS will occur regardless of the animal’s behavior.
UCS
(food, electric shock).
UCR
(salivation, limb movement. Inborn.).
CS
light, tone
CR
salivation, limb movement. Acquired through learning.).
Order of stimulus
critical to learning. Forward pairing (CS before UCS) is best, simultaneous pairing is okay, backward pairing (UCS before CS) is poor (if any). This makes sense because classical conditioning should prepare the body for a response.
Acquisition of a Classical Conditioning Response
learning is gradual. A reinforced trial is simply a pairing of CS and UCS. The strength of the CR is determined by omitting the UCS (unreinforced trial) and looking for a CR. A growth curve tends to look like gradual rise and then leveling off, although it depends of the specific CS and UCS.
reinforced trial
simply a pairing of CS and UCS
unreinforced trial)
strength of the CR is determined by omitting the UCS (unreinforced trial) and looking for a CR
Extinction of Response
when CS and UCS pairing stops. After extinction the decline of response does not begin immediately. When pairing ceases the CR’s strength eventually declines to zero. But it is not “gone” because some of the response returns (spontaneous recovery).
Reacquisition/reconditioning
even when response is abolished (no spontaneous recovery remains), some learning remains because the reacquisition curves moves much faster than the original one. Extinction teaches us that the pairing is no longer valid, but if it becomes valid again, there is a basis for quick relearning!
Generalization
CR will be linked not only to CS but also to similar stimuli (tone of 1000 as CS will also create response to 800, 1200, etc), which is important because stimuli in nature vary from sample to sample. Generalization gradient is the pattern/graph. CS+ and CS- are opposite responses. CS- is an inhibitor that inhibits response elicited by CS+.But,
Discrimination
shows that one can learn not to respond to a generalized stimulus if there is no pairing. This feature has adaptive value.
Conditioned emotional response (CER)-
like for fear.
Compensatory reaction
(homeostasis, think insulin/addition, inhibition)
third-order conditioning and Pavlov
Pavlov thought that complex learned behavior could be explained by a long chain of CRs (food to tone, then tone to light, etc), but third-order conditioning is nearly impossible, so complex learned behavior must involve something else as well.
Instrumental/operant learning
allows an animal to “interface” with the external environment to attain certain goals. Creates connections between actions and their consequences. Reward depends on proper response. Act outcome representation- an animal learns to associate an act with its outcome.
Act outcome representation
an animal learns to associate an act with its outcome.
Behaviorism
you shouldn’t discuss “the mind” but rather behavior because the mind is not observable. (This idea has since gone out of fashion.)
Thorndike’s puzzle box-
cats were deprived of food and then put into a puzzle box once a day. Response measure (time needed to get out of box in seconds- latency). Because of jagged, clear, gradual trend, Thorndike assumed that animals were learning by trial and error, not by “thinking.” (Behaviorist view.)
Thorndike’s Law of Effect-
the sequences (effect) of a response determines whether the tendency of the stimulus to produce that response again is strengthened or weakened. A reward increases the probability of the response, while a punishment decreases it. Response rate = number of responses per unit of time. What had been learned was a link between the stimulus situation and the response (S-R bond). Rewards strengthen (stamp) S-R bond. Adaptive nature of such processes- only the fittest responses survive (Darwinian!).
Skinner-
also a behaviorist, believed you could shape a child into anything bases on learning experiences. Against idea of “free will.”
Skinner box/lever box
give reward in small doses, multiple trials per day. Animal said to “emit” an operant response that operates on the environment.
Present stimulus Remove stimulus
Pleasant stimulus positive reinforcement (up) omission training (down)
Unpleasant stimulus punishment (down) negative reinforcement (up)
Also known as appetitive and aversive stimuli.
Primary reinforcers/secondary (conditional) reinforcers
Primary reinforcers- reinforcers that are “built into” an animal (like food). However, one can arrange for what is normally a neutral stimulus to become reinforcing (secondary reinforcement). Also known as conditional reinforcers. Tokens, smiles, etc get responses because they lead to primary reinforcers. With tone, bar press, and food, tone will become a secondary reinforcer. Secondary reinforcement greatly extends generality. Very convenient as opposed to primary reinforcers. This fact allows for shaping behavior (and successive approximations). This idea is closely linked to classical conditioning.
Partial reinforcement
response is only sometimes reinforced (like fishing).
partial reinforcement effect
it is easier to get more behavior by giving reinforcements less frequently (known as the partial reinforcement effect). Also, slower extinction.
Schedules of reinforcement-
Cumulative responses- running sum of the response through time, displayed by cumulative record (graph). Types: continuous reinforcement (low rate of responses). Fixed interval (reward given after certain amount of time, scallops because animal knows that no reward will come after it has just been given one). Variable interval (after average amount of time, higher rate of response, no real pauses). Fixed ratio (reward after certain number of responses, pause after reward has been given). Variable ratio (after average amount of responses, very high responses rate, no real pauses).
Thorndike
believed that learning could be reduced to associations between specific stimuli and muscle twitches. But, think of bell/electrolode example: if finger is turned upside down, it will perform a different evasive procedure after hearing bell to avoid the shock. Remove finger is the message, NOT move it forward.
Cognition-
viewing learning as the acquisition of knowledge, not the change in behavior.
latent (hidden) learning.
Tolman argued that animals learn from strategies and relations, not mindless/robot S-R links. Tolman disputed the behaviorist approach (animals learn only if presented with a reward) by pulling rats through a maze on a rat mobile with no reward present. Rats made a cognitive map (mental representation of spatial layout) even without a reward. When on 12th day the rats were presented with a reward (after no previous signs of learning the maze), they performed just as well as the control group (which was rewarded all along). Idea of latent (hidden) learning.
Distinction between learning and performance
rewards are not just needed for learning but are needed for the performance that reveals the learning. Animals are naturally curious/inclined to learn cognitive maps even without a reward.
belongingness and preparedness
Not all associations are equally likely for each species (rejection of equipotentiality principal). Some constraints are present. Idea of belongingness and preparedness. Rats respond to taste but not color (one-trial learning), birds to color but not taste. Learned taste aversion can be very long-lasting. Generalizations are species-specific.
insight learning
Even Tolman didn’t believe that learning involved rational analysis. But, Kohler found evidence of insight learning in apes. Trial and error learning is ruled out. Apes, after a period of exploration and frustration, suddenly (ah-ha!) solve the puzzle and put the two sticks together to reach the food. Learning meets thinking!
Contiguity reconsidered
two things (S-S or S-R) are contiguous if they occur at the same time (temporal contiguity). But, there are more to associations than just contiguity. If light is flashed before tone for food, only the light becomes a secondary reinforcer. This is because the tone is contiguous but redundant. If light is flashed before tone but light is also presented when it’s not connected to food, only the tone will become a secondary reinforcer (the light isn’t reliable). A better term is contingency, which requires non-redundancy and reliability.
contingency,
requires non-redundancy and reliability.
Fear and anxiety
Fear- elicited by something specific. Anxiety (chronic, objectless, worse than fear).
Learned helplessness

Neural plasticity-
an acquired sense that one cannot control one’s environment.

capacity of neurons to change the way they function to account for learning (new synapses, etc.).

Deck Info

45

permalink