Upgrade to remove ads
Only $35.99/year

Learning and Motivation- chapter 8

Terms in this set (49)

what do mechanisms of stimulus control allow you to learn?
-mechanisms of stimulus control allow you to learn to discriminate situations in which behaviors are rewarded from situations in which they are not
Bond and Kamil- what did they demonstrate with moths?
Bond + Kamil → demonstrated that stimulus control processes can affect both a predator + its prey
When moths more conspicuous than others, blue jays tended to detect + peck at the conspicuous ones + miss the more hidden/cryptic moths → evolutionary terms, payoff for moths to be cryptic: cryptic moths that escape detection are more likely to have offspring in the next generation
moths that the blue jays detected and pecked in one session were...
moths evolved to become more what?
-moths that the blue jays detected + pecked in one session eliminated from pop that was presented in next session → moths that not pecked survived + represented as "offspring" that differed only a little from the parents → blue jays introduced natural selection, + moth population changed

Moths evolved to become more cryptic + gradually came to look different from one another (polymorphisms)
which moths were more likely to survive?
-moths that were different than their neighbors were more likely to survive than moths that were not different (psychology of stimulus control affected behavior of predator + evolution of prey)
-animal cognition- study of cognitive processes in animals
categorization and discrimination:
what is the Herrnstein, Loveland, and Carble experiment with pigeons in a box? when they are shown a series of color slides, what is SD? when are they reinforced?
-pigeon box w screen that allowed them to show the birds a long series of color slides → birds viewed each slide for 30s → in the presence of half slides, pecking key reinforced. In presence of other slides, pecking not reinforced
-bird should have learned to peck only when a positive slide (SD) was shown on the screen
-each bird tested + trained on 1700 slides → for some birds, slides that signaled the pecking would be reinforced contained an image of a tree, + slides that signaled no food did not contain a tree

-slides taken all 4 seasons → some slides only contained parts of trees, diff angles, etc
-despite variable tree images, birds learned to discriminate tree slides form other slides (pecked more during tree slides)
when the birds were tested with new slides that they had never seen before, they responded mainly if...
the new slide contained a tree → generalized to new examples, learned to categorize pictures as those w trees or w no trees
-other pigeons trained + tested with other categories (ex: with water) → learn to categorize accurately + responded to new pictures
-categories complex + open-ended (polymorphous stimulus sets, very common in natural world)
according to wasserman, what do pigeons learn to discriminate? what is an experiment surrounding this?
-pigeons learn to discriminate between four categories at the same time → pigeons shown images on a small computer screen + then required to peck one of 4 keys located near the corners of the screen → to receive reinforcer, pigeon must peck one of the keys, depending on whether the image on the screen is a cat, car, flower, or chair → must report the "name" of the image by pecking a particular key
-birds learn this task rapidly + accurately → pictures that make up each category appear to be similar but still discriminably different to the birds
when pigeons tested with new images that they've never seen before, how did htey respond?

when birds were trained with a large set of pictures that were never repeated, what were they able to do?
-when pigeons tested with new images they've never seen before, respond well above chance, although performance is a bit worse than with images the birds have specifically been trained to see
exp - birds trained with a large set of pictures that were never repeated → even when exemplars never repeated, pigeons learned to discriminate between cars, chairs, flowers, + people
when wasserman trained pigeons to discriminate between color pictures from different categories, what occurred?
-Wasserman: trained pigeons to discriminate between color pictures from 16 different categories → in each training session, birds were reinforced for ecking a distinct image that was associated with each category

-on each trial, color picture presented at the center of the screen, and the birds were reinforced for pecking at an image that "labeled" the category instead of a randomly-chosen image that labeled another one → birds learned to label pictures that were examples of each of the 16 categories + later showed good performance when tested with new pictures that they'd never seen before
within this kind of experiment, what is important, and why?
memories for specific photographs → pigeons have capacity to remember over 800 photographs
-to make the categorization of stimuli more accurate, best if the bird has experience with many different examples during training
what did Bhatt do with pigeons in regards to examples in each category? the more samples used in training, the....
If the goal is to increase performance in new situations, what is best?
-Bhatt: gave diff groups of pigeons either 1 example of each category, 4 examples, or 12 examples of each → task involving only 1 example of each category was easiest for pigeons ot learn → transfer to new pictures was bad though... the more samples used in training the better the transfer to new images

-if goal is to increase performance in new situations, more examples in training the better, even though the learning may be a little difficult
what does feature theory accept?
when pigeon is discriminating images of cats from non-cats, how does feature-theory explain this?
Feature theory: accepts the idea that all pictures contain many different features and that the bird must learn which features signal reinforcement and which do not

Ex: pigeon discriminating images of cats from non-cats → one picture of cat might contain several features (picture=compound stimulus) → when compound paired w reinforcer, each feature might be associated with it a little (associative strength of each increases a bit each trial) → associative strength of cues that repeated over trials increases → on negative trial, slide might show a dog rather than cat: may have another set of features... each of these cues would receive a decrement in associative strength + possibly inhibition
according to feature theory, after a number of trials, what happens to the features?
-after a number of trials, several features would have associative strength: negative (inhibitory) or positive (excitatory)
-To test feature theory: present stimuli with features that experimenter can manipulate easily
what did Castro and Wasserman teach pigeons to do? on each trial, what was presented?
showed pigeons series of displays on touchscreen, differed from trial to trial
-each display had distinctive feature at each corner, but features presented in different corners from trial to trial → two features always indicated that display a member of Category A, and two other features indicated from category B
-on each trial, display presented + pigeon had to peck display a few times → then, 2 response buttons: responses to button A reinforced if display from category A, and pecks to button B reinforced if from Category B

-birds learned to categorize the displays very well → also directed early pecks at relevant features in the corners
when huber and Lenz showed pigeons images of faces, what could each dimension have?
when they showed the birds pictures of a number of faces made up of different combinations of the features, what happened?
showed pigeons images of faces (varied w: size of forehead, space between the eyes, length of nose, size of chin below mouth) → each dimension could have one of three values (-1 → 0 → +1)
-then showed the birds pictures of a number of faces made up of different combinations of the features → pecking reinforced if the face contained features with values that summed to more than 0

-birds learned discrimination → pigeons learned ab relevant features + respond according to their value (peck speed)
with some sets of stimuli, pigeons appear to respond as if....
what is the prototype theory?
-with some sets of stimuli, pigeons appear to respond as if they have learned a prototype of the stimuli in the category → birds respond to stimuli to the extent that the stimuli were similar to an "average" rather than the sum of the features (prototype theory)
-prototype theory: exposure to the different types of trials results in the formation of a prototype
-problem: results suggesting that learning of a prototype can often be explained by feature theory
what is the exemplar theory?

what stimuli do animals learn about? how do they respond to new stimuli ?
Exemplar theory: animals might learn to categorize in the simplest possible way → bird may learn + remember each picture or image presented to it in the training phase → when novel stimuli tested, bird might generalize from all exemplars + respond to the extent that the new picture is similar to a picture that was reinforced before
-animals learn ab whole configurations of stimuli + then respond to new stimuli according to how well they generalize to them
what is a generalization gradient?

in Guttman and Kalish's experiment, where pecking is reinforced at a simple key illuminated with a specific color, what happened after training when extinction occurred?
-behavior that has been reinforced in the presence of one stimulus generalizes to similar stimuli

Guttman + Kalish exp: reinforced pecking at a simple key illuminated with specific color → after training for several days, birds tested with trials in which the color of the key light was varied in many steps between green, yellow, orange, red → tests conducted in extinction (none of new tests colors reinforced) → birds pecked at the new colors if their wavelength was similar enough to the original value
in this experiment, what was the stimulus generalization gradient? what does the steepness of the generalization gradient indicate?
what does the shape of the gradient indicate?
-stimulus generalization gradient: responding to a new stimulus depends on its similarity to a stimulus that has already been reinforced
-steepness of generalization gradient indicates how much the responding actually depends on a particular stimulus dimension
-shape of gradient not automatically controlled by cues present when an operant is reinforced → shape of gradient can be affected by learning
-another stimulus has to be non reinforced (SD must be discriminated from Stheta)
what is an example of this with pigeons, where a 1000 Hz tone was manipulated in the background?

1 condition? other condition? final condition?
-Ex: reinforce pigeons for pecking a key that was white when illuminated → manipulated presence of 1000 Hz tone that could be sounded in the background

-1 conditioning: pecking key reinforced if tone was on but not reinforced if tone off → birds eventually responded mainly when tone on
-other condition: birds learned to discriminate between 2 tones: pecking reinforced when 1000 Hz tone on, + not reinforced when slightly lower tone on
-final condition: birds received no discrimination training → 1000 Hz tone always on, + birds reinforced whenever pecked illuminated key
when the birds had not received explicit discrimination training, what was generalization like?
when they were reinforced while the tone was on and not when it was off, what occurred?
when was the sharpest reinforcement gradient seen?
-when birds had not received explicit discrimination training, generalization flat (equal responding to all tested tones, no evidence of stimulus control by tone in the situation)
-when birds reinforced while the 1000 Hz tone was on but not reinforced when off, stimulus generalization gradient
-sharpest reinforcement gradient by group that had learned to discriminate 1000 Hz from 950 Hz (max responding near 1000 Hz and nothing to other stimuli)
how does the process of inhibition affect the steepness of the generalization gradient?

in an experiment where there were two groups of pigeons, where they were taught to discriminate a plain white key with a vertical black bar on it, how did the groups differ?
in the final generalization test, what were both groups presented with?
Exp: 2 groups of pigeons, taught to discriminate plain white key from white key with vertical black bar on it
-for one group, pecking reinforced when the vertical black bar was present, but not when no black bar
-other group: opposite treatment, pecking reinforced when key only white

-both groups learned to peck when appropriate
-final generalization test: both groups presented w usual vertical black bar on key, but on different trials presented at different angles
what were the results?
-group that had been trained to peck in the presence of vertical black bar showed another gradient (clear peak w vertical, + responding declined as angle of bar changed)
-for birds where black bar signalled no reinforcement: birds responded little to black bar, but pecking increased as angle of the bar different (inhibition develops to a stimulus that signals nonreinforcement of an operant response) → feature-negative discrimination learning
What did spence suggest about S+? when will it receive excitatory conditioning?
when will S- receive some inhibitory conditioning?
-Spence: suggested that an S+ will receive some excitatory conditioning when its reinforced + the excitation will generalize from it to other similar stimuli
-S- will receive some inhibitory conditioning when non reinforced, + inhibitory conditioning will generalize
-gradients consistent with Spence's theory
what does reinforcement of a 550 nm light condition? what generalizes to other stimuli?
what about non-reinforcement?
what did spence think in regards to responding to new stimuli?
-reinforcement of 550 nm key light will condition excitation to 550 nm, and excitation generalizes to other similar stimuli
-nonreinforcement of similar key light color conditions inhibition to it → generalizes

-suppose that test responding to all stimuli surrounding S+ and S- → Spence: assumed that responding to the new stimuli would depend on how much inhibition + excitation generalize to the new stimuli from S+ and S- → inhibition would subtract from excitation
what happens when already see S-, but then S+ given?
-what actually see: bc of the inhibition generalizing from S-, theory does not predict most responding at the trained S+ (predicts even more responding w stimuli farther away from S-) → peak of generalization gradient should shift away from S- to a stimulus that was never reinforced
what were hansons three groups of pigeons?
Hanson: one group pigeons reinforced for pecking a 550 nm light + not reinforced for pecking an extremely similar 555 nm key light
-second group: received discrimination training except that th enonreinforced S- was 590nm
-third group: received reinforcement with 550 nm (control)
when the groups were given discrimination training, where did they see more responding ? when was this especially true?
-groups given discrimination trianing showed more responding to 540 nm than 550 nm → esp true of pigeons given 555 nm, which should have generalized most strongly to 550 nm → phenomenon where the peak of generalization gradient moves away from S-=peak shift
what do peak shifts suggest?
-peak shift:

-suggests that the highest level of responding can somewhat paradoxically occur to stimuli that have never been directly reinforced
-suggests possible role of inhibition
-might have implications for evolution (may explain why elaborate characteristics/coloration might evolve in members of one sex)
suppose you train an animal so that a relatively bright stimulus (S+) reinforced and relatively dark stimulus (S-) not --> what could this lead to? what does the strength of responding to a given stimulus depend on?
-suppose train an animal so that a relatively bright stimulus (S+) reinforced + relatively dark stimulus (S-) not → could lead to gradients of excitation + inhibition around S+ and S-
-strength of responding to a given stimulus will depend on the difference in the amount of excitation + inhibition that generalize to it
if given the choice between S+ and S-, will chose...

suppose that an animal is given choice between S+ and a brighter stimulus S'- what do gradients suggest?
-if given choice between S+ and S-, will choose S+
-suppose that animal given choice between S+ and brighter stimulus S' → gradients suggest that animal should actually prefer brighter stimulus → transposition
transposition- what does it suggest?
what was the problem for spence?
-transposition: suggests that animals respond to relationship between two stimuli rather than their absolute properties
-problem for spence: transposition can still occur under conditions that can only be explained by relational learning
how to organisms generalize? what does it interact with?
what does peak shift phenomenon suggest?
-organisms generalize from one stimulus to other stimuli that are similar → process interacts with discrimination learning → peak shift phenomenon suggests that generalization gradients can influence the results of explicit discrimination training → discrimination learning can influence the extend to which animals generalize from stimulus to stimulus


-discrimination + generalization processes always interacting
in perceptual learning, what can fine discriminations result from?
fine discriminations can result from differential reinforcement of the different stimuli → further facilitated by mere exposure to the similar stimuli without any differential reinforcement
Gibson + walk experiment?
compared to rats that had no exposure to rectangle and triangle, rats exposed to these cues were....

perceptual learning explanation?
hung large metal rectangle + triangle on wall of home cage of group of rats for several weeks

-compared to rats that had no exposure to rectangle+triangle, rats exposed to these cues were quicker to discriminate between then when one shape used as an S+ and one sued as an S- in a discrimination exp

-perceptual learning: simple exposure made the 2 different shapes easier to discriminate → since neither stimulus was reinforced in initial phase, rules for discrimination learning dont easily apply → need differential reinforcement of one stimulus over the other to achieve differences in excitation or inhibition
what is one idea about the results of this study?
-one idea: exposure somehow led the animals to differentiate the stimuli (mechanisms=vague)
-theories of conditioning also challenged by perceptual learning bc usually predict that non reinforced exposure to stimulus should reduce conditionability (LI) → rate of learning could be quite slow, but difference in responding to S+ and S- might develop relatively quickly
According to McLaren and Mackintosh, when will organisms generalize between two stimuli + have trouble discriminating between them?

If stimulus A is presented, what does it present?
-organisms will generalize between two stimuli + have trouble discriminating between them to the extent that they share common features → each stimulus has unique elements + some common overlapping elements (ex: stimulus A and B, overlapping region=X)

-if stimulus A presented, actually presents a combination of unique + shared elements (A and X) → if B presented, B and X
if stimulus A was paired with a reinforcer, A and X would both acquire...

pre-exposure to stimuli A and B would involve...
-if stimulus A paired with reinforcer, A and X would both acquire associative strength → would respond to stimulus B (X) and show generalization bc the common element (X) received some conditioning

-preexposure to stimuli A and B would involve exposure to two compounds Ax and Bx → bc neither A or B reinforced in preexposure phase, AX and BX would be latenly inhibited (be more difficult to condition in next phase)
if there are 4 pre-exposures to stimulus A and 4 pre exposures to B, what receive more latent inhibition?
-if 4 pre exposures to stimulus A and 4 pre exposures to B, 4 pre exposures to A, 4 to B, and 8 to X → common elements receive more latent inhibition than unique elements (common elements=especially difficult to condition)
In Mackintosh, Kaye, and Bennett's study, where rats drink sucrose and saline solutions that could be mixed with lemon, what occurred?
-rats drink saline + sucrose solns that could be mixed with lemon → saline, sucrose, + lemon=a, b, and x
-if rats received saline (A) w illness + tested w sucrose (B), no generalization

-rats received conditioning with saline-lemon compound (AX) and sucrose-lemon compound (X) → if rats first received pre exposures to AX and BX, much less generalization between them → pre exposures to the compound stimuli made them easier to discriminate

-group 3: received same pre exposures to X, but alone → after conditioning with AX, showed little generalization from AX to BX → preexposure to common element (X)=sufficient to make it easier to discriminate the compounds
According to McLaren, during exposures to complex cues, what do animals learn to associate? what is the consequence of this?
-McLaren: during exposures to complex cues, animals learn to associate various elements → AX; might learn to associate A with C, and during BX, learn to associate B with X → consequence: because x associated with b, x will retrieve b when rat being exposed to ax (will "expect" b during ax) → rat adjusts its expectations accordingly, learns to expect "no b" in presence of a → also learns to expect "no a" in presence of b... leads to inhibitory associations between a and b
-inhibition between unique elements=second mechanism behind perceptual learning
if the stimuli are especially complex, an organism might be able to sample only a small portion of the elements on any given trial if....
-third mechanism: if stimuli are especially complex, an organism might be able to sample only a small portion of the elements on any given trial → if associate those elements on each trial + then sample + associate some new elements on each successive trial, whole network of interconnected elements can be built up over trials → associations between different elements create a "unitized" representation of the stimulus → allows subset of features sampled on any given trial to activate the entire representation (pattern completion)
if exposure to the different features is to facilitate conditioning, what must happen to the different features?
-different features must be experienced at the same time (rather than separately) if exposure to the different features is to facilitate conditioning → form representations of complex stimuli when exposed to them, so unitization=third possible mechanism that leads to perceptual learning
2 stimuli can become more discriminable if ....
-2 stimuli can become more discriminable if merely exposed to them without differential reinforcement → discrimination might occur if stimuli composed of many different elements that can be latelnyl inhibited or interassociated → laws of learning are happening everywhere
if two stimuli each associated with a common third stimulus, they will begin to .....

in the Honey and Hall experiment, 2 auditory cues are paired with food --> then, N is paired with electric shock. What occurs during the experiment, and what are the results?
-Honey + Hall: paired 2 auditory cues (A and N) with food → common associated w food expected to increase the generalization between A and N
-N paired with electric shock on several trials to N aroused fear
-then tested fear to A + to a new auditory cue, B
-results: fear generalization from N to A seen, but not to B, and no generalization in a control group that didn't have A and N paired with food
when two stimuli are associated with something in common, tend to treat them as if they are....
-when 2 stimuli associated with something in common, tend to treat them as if they are more equal (acquired equivalence/mediated generalization - generalization between A and B is "mediated" by their association w a common element, + A + B retrieve a common representation)
when pigeons are run in experiments using variations on the matching-to-sample procedure, what happens?
Ex: pigeons can be run in experiments using variations on the matching-to-sample procedure → on each trial, after initial "start" cue, "sample" stimulus presented on central key → immediately after this stimulus, bird shown 2 "comparison" stimuli on the side keys → bird reinforced for pecking the comparison key that matches the sample → always involves conditional discrimination bc either comparison stimulus is reinforced on half trials, but the one that will be reinforced on any particular trial depends on the sample cue that came before it

-when different samples signal that same comparison stimulus is correct, different samples treated as equivalent
what is the first phase (acquired equivalence and mediated generalization)? final test?
Ex: pigeons can be run in experiments using variations on the matching-to-sample procedure → on each trial, after initial "start" cue, "sample" stimulus presented on central key → immediately after this stimulus, bird shown 2 "comparison" stimuli on the side keys → bird reinforced for pecking the comparison key that matches the sample → always involves conditional discrimination bc either comparison stimulus is reinforced on half trials, but the one that will be reinforced on any particular trial depends on the sample cue that came before it

-when different samples signal that same comparison stimulus is correct, different samples treated as equivalent
what do you respond to in regards to psychological similarities?

what does Lea think is an important step in demonstrating true conceptualization?

when can generalization occur?
-respond to psychological similarities created by associations with response to common key
Lea: ability to categorize stimuli on a basis that transcends physical appearance=imp step in demonstrating true "conceptualization" in animals ("categorization by association") → subordinate categories

-generalization can occur between cues that have little in the way of physical similarity → common associations="glue" that ives different cues a psychological similarity of equivalence
Other sets by this creator