Learning Psychology 3510 Fall 2016 Professor Delamater Selected Special Topics in the Study of Comparative Cognition Episodic memory in nonhuman animals Interval Timing Serial Order learning ID: 542227
Download Presentation The PPT/PDF document "Lecture 23: Animal Cognition II (Specia..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Lecture 23: Animal Cognition II (Special Topics)
Learning, Psychology 3510
Fall, 2018
Professor DelamaterSlide2
Selected Special Topics in the Study of Comparative Cognition
Episodic memory in non-human animals
Interval Timing
Serial Order learning
Categorization and Concept LearningSlide3
Episodic Memory: Food Caching Birds
Kamil
and
Balda
(1985) study
Clayton and Dickinson (1999) study
Clark’s nutcrackers live in Alpine regions of the western US.They are known for their uncanny ability to retrieve cached food in the wild. They store seeds from pine cones during the summer and fall, and then retrieve them in the winter and spring.They can store up to 33,000 seeds and recover several thousand of them at some later time.Can they do this because they remember well, or because they can spot markings of where they’ve been? Or just randomly run into food they’ve or others have previously stored?
Kamil
and
Balda
(1985) studySlide4
Episodic Memory: Food Caching Birds
Kamil
and
Balda
(1985) study
Clayton and Dickinson (1999) study
Kamil and Balda forced the birds to store up to 18 seeds in a laboratory environment with 225 storage holes. However, only 18 were available at any given time.Then, after some delay period, they were allowed to retrieve their stored caches.But, during this retrieval test, all 225 storage holes were accessible.
The birds could recover their caches 2-4 times greater than would be expected by chance.
Furthermore, they could remember these after as long as a 285 day delay.Slide5
Episodic Memory: Food Caching Birds
Kamil
and
Balda
(1985) study – This study shows an amazing memory ability.
Clayton and Dickinson (1999) study – This study shows some evidence for episodic
memory in birds (Scrub Jays).What is “episodic” memory? Is it uniquely human? a. Usual answer is that it is a recollection of a particular event that took placein a particular location at a particular time. In essence, you reexperience theepisode. b. Aside from the conscious aspects of this memory, it can be characterizedAs a memory containing “what,” “where,” and “when” information. c. Clayton and Dickinson (1999) asked if scrub jays had this kind of memory.Slide6
Episodic Memory: Food Caching Birds
Clayton and Dickinson (1999) study
Birds were asked to cache different types of food (peanuts and mealworms) into ice
cube trays that had sand in them and were surrounded by
lego
block structures.
Then they were asked to retrieve them after 4 or 124 hour delays.Some birds learned, though, that the meal worms deteriorated after 124 hours, whereas for others the bad mealworms were replenished with fresh ones.Slide7
Episodic Memory: Food Caching Birds
Clayton and Dickinson (1999) study
Since the jays preferred mealworms to peanuts, the Replenish group always retrieved
more mealworms (W) than peanuts (P) during the 4 and 124 hour tests.
The birds that learned that mealworms decay after 124 hours preferred searching for the peanuts to the mealworms after 124 hours, but worms to peanuts after 4 hours.
This means that the birds knew after 124 hours what food was stored, as well as where and when it was stored there.
This suggests that the scrub jays, like humans, may have episodic memories.Slide8
Interval Timing: Methods & Properties
Duration Estimation (symbolic matching to sample task….)
In the duration estimation task, the subject judges whether the first stimulus (the sample stimulus) is short or long.
If it is short, then choosing the left response is reinforced with food, but if it is long, then choosing the right response is reinforced with food.
Tone
Tone
Choose Response 1 if short
Choose Response 2 if long
Sample Stimulus ChoiceSlide9
Interval Timing: Methods & Properties
Duration Estimation (symbolic matching to sample task….)
Peak Procedure (duration production)
In the peak procedure, a stimulus comes on and signals reward opportunity after a fixed interval (e.g., FI 20 s).
But on
nonreinforced
probe trials, the duration of the trial is extended and no reward can be earned.It is on these trials that we assess responding as a function of time.
Response
S
Training trials
Fixed Interval 20 s
S
Nonreinforced
Probe trials
80-s S duration
Reward
ResponseSlide10
Interval Timing: Methods & Properties
Duration Estimation (symbolic matching to sample task….)
Peak Procedure (duration production)
In this case one discriminative stimulus signaled FI 20 reward, while a second signaled FI 40.
The rats lever press rate on probe trials peaked at 20 and 40 s, respectively, in the presence of the two stimuli.
This indicates that they have learned about reward time.
Also, note that the variability (spread) of the two timing functions differ.The amount of variation increases with the mean of the interval being estimated.
Roberts (1981) study
In fact, responding is related to the proportion of time elapsed, independent of the value of the interval being judged. This is known as the scalar invariance property of interval timing.Slide11
Interval Timing: Theories
Scalar Expectancy Theory (Gibbon & Church, 1984)
Behavioral Theory of Time (Killeen &
Fetterman
, 1993)
Multiple Oscillator Theory (Church & Broadbent, 1990)
There are clock, memory, and decision processes in the scalar expectancy model.Clock generates counts in an accumulator.Memory: Running count vs reinforced memory count (at time of reward).
Decision: Compares running value of elapsed time in working memory to the stored reinforced time in reference memory.
Comparator rule shows how behavior will be proportionally related to the value of the reinforced time in reference memory.
Gibbon and Church, 1984
Comparator:
(N* - N)
< b (threshold)
N*Slide12
Interval Timing: Theories
Scalar Expectancy Theory (Gibbon & Church, 1984)
Behavioral Theory of Time (Killeen &
Fetterman
, 1993)
Multiple Oscillator Theory (Church & Broadbent, 1990)
In the Multiple Oscillator theory, there are different oscillators that run concurrently.Each oscillator has a different periodicity.Each moment in time has a different pattern of activation across the different oscillators.These different patterns of activation are thought to code different temporal intervals.
Church and Broadbent, 1990
Time 1 Time 2Slide13
Serial Order Learning
Learn a sequence of responses: 123 234 345 456 567 678 (Fountain, et al, 2012)
Do the rats learn an internal representation of the sequence or do they learn
a chain of S-R associations?
They learn a structured sequence better than a non-structured one.
But that could be understood in terms of easier associative learning….Slide14
Serial Order Learning
Learn a sequence of responses: 123 234 345 456 567 678 (Fountain, et al, 2012)
Do the rats learn an internal representation of the sequence or do they learn
a chain of S-R associations?
3. Terrace, Son, and Brannon (2003) study
Monkeys first learned 4, 7-item sequences:
A1 – A2 – A3 – A4 – A5 – A6 – A7 B1 – B2 – B3 – B4 – B5 – B6 – B7 C1 – C2 – C3 – C4 – C5 – C6 – C7 D1 – D2 – D3 – D4 – D5 – D6 – D7These stimuli were different pictures taken from several different categories.
Then the monkeys were given a set of 2-item probe tests consisting of different
stimuli from the different sets. They asked whether the monkeys could
accurately select the first item.
A3 – C5, B2 – D4, C5 – D7, D3 – A4,
etc
The monkeys DID choose the “earlier” item of each sequence even though these
particular items were never trained together.
It looks like the monkeys had constructed a representation of the order of each sequence, and this order guided their choices, not a series of S-R associations.Slide15
Categorization & Concept Learning
Classic Herrnstein, Loveland, and Cable (1976) study
Pigeons learn to categorize pictures of people
vs
plants.
A control group is trained to arbitrarily categorize different sets of stimuli.
The individual stimuli chosen for this study are taken from a large number of items (somewhere on the order of 100 or more).
Then, during a test phase the animals are tested with NEW exemplars (ones they never saw before), and were asked to choose to respond.
The birds learning the Categorization task did so fairly well, with some loss in accuracy.
Later work has shown that greater generalization occurs in these tests as the number of exemplars in the training set increases.Slide16
Categorization & Concept Learning
Grainger, et al (2012) study with baboons
Baboons first learned to categorize English words
vs
non-words (up to over 300 exemplars of each)
Then they were tested with new words and non-words they
hadnt seen before.They’re performance was pretty good, as can be seen on the graph to the right.Slide17
Categorization & Concept Learning: Mechanisms of Concept Learning
Aust
and Huber (2003) study with pigeons
The birds first learned to classify pictures
of people
vs
no peopleThen they were tested with new exemplars of people that were in their original form or were rearranged in various ways.The birds responded most when the new
people were in their original form, but
they responded more to rearranged
people than to slides with no people in them.
This suggests that the birds responded at least partly on the basis of the specific features making up the original slides.
This is the Feature Theory of Concept learning.
Notice how this is rather similar to the
Rescorla
-Wagner theory of simple conditioning.Slide18
Categorization & Concept Learning: Mechanisms of Concept Learning
2. Wright and Katz (2007) study with pigeons, rhesus monkeys, capuchin monkeys
This was a Same/Different concept learning task
Do these species learn “abstract” concepts?
If the two objects on a trial are the same, then
touching the lower picture is correct.
If the two objects are different, then touching the button to the right of the lower picture is correct.After the basic learning phase, then the animals
are given tests with new exemplars.
The basic findings are that generalization to new exemplars increases as the number of training exemplars increases.
This result occurs in all species suggesting that all species may learn in similar ways.
Perhaps they are abstracting some general notion of sameness, and responding on the basis of this representation. Further work is needed to clarify the nature of that representation.