nf1207: one on learning next week Dr namex will be starting the final section 
of this psychobiology course but this is the last one of mine and as i 
indicated in my last lecture up till now would you stop talking please thank 
you up till now i've been talking about animals learning individually and in 
this last lecture i'm going to talk a bit about what animals learn in the 
company of other animals in particular whether they learn from observing other 
animals and if so what er the terminology i'll be using is that i'll be talking 
about quite a lot of experiments where some animals are observers that is they 
watch other animals performing and and other animals are the demonstrators or 
models that is they perform something or behave in a certain way and we want to 
know what the observers learn from watching the demonstrators or models well 
what might they learn basically the two kinds of things that they might learn 
correspond to the two kind of individual sorts of associative 
learning that i've been talking about in these lectures so far for example 
animals might learn from one another the significance of stimuli what foods are 
good or bad to eat what things are predators these are the kind of things that 
we know they can learn individually through Pavlovian conditioning but maybe 
they can learn it from watching other animals what other animal eat what other 
animals avoid as well the other kind of new learning i talked about in my last 
lecture is learning new actions for example learning how to get at or handle 
food this is the sort of thing that animals can learn individually through 
instrumental conditioning as i discussed in my last lecture perhaps they can 
also learn it to some extent by watching other animals and the two the sections 
of this lecture refer to experiments in which these two potential kinds of 
social learning are studied before talking about experiments i want to talk a 
bit about the kind of naturalistic 
data which suggest to people that this observational or social learning goes on 
that animals learn things from other animals in their social group or culture 
which they would not learn in other situations well there is lots of apparent 
evidence which has been interpreted as evidence of social learning for example 
if you look at different colonies or troops of chimps the same species but 
living in different parts of Africa you find that different ch-, troops show 
different behaviours for example some troops of chimps build nests in trees 
others don't is it possible that the troops that show this behaviour have 
learned it by joint learning by copying each other whereas the troops that 
don't show it have simply not had any demonstrators around so haven't had the 
opportunity to learn similarly certain specialized kinds of food getting 
behaviour there's a behaviour i-, a well known behaviour observed in chimps by 
Jane van Lawick-Goodall er which is termite fishing where a 
skilled chimp chimps feed off termites which are little ant-like creatures 
which live in mounds the termites have holes through which they enter and exit 
from the mound and chimps in some colonies have learned to get twigs or 
branches strip off the leaves from these twigs so they are nice and thin and 
pointed and poke the twigs down the holes into the termite mound sit and wait 
eventually termites will crawl on to the stick at which point the chimp whips 
the stick out and eats the termites off the stick now this is a quite a skilled 
behaviour but it's a very cunning way of getting termites out of mounds and 
it's shown by some chimps and not others and more to the point young chimps 
have been observed watching their mothers do this and the question arises to 
what extent do chimps learn this skill by watching their mothers er another 
example of a similar kind which i mentioned in my last lecture is illustrated 
in this overhead it shows black rats i said in 
my last lecture black squirrel that was a mistake it's black rats feeding off 
pine cones you remember i said that black rats get quite skilled at stripping 
the exterior off pine cones so they can get at the kernels in the middle well 
adult rats of course do this skilfully and in the wild it's very often the case 
that young rats feed together with their parents watch them while they're 
feeding off pine cones and attempt to feed off pine cones at thems-, themselves 
at the same time so the question is to what extent are the young rats learning 
this skill by watching the black rats well you might say what are the 
alternatives the answer is there are lots of alternatives one possibility is 
just that young rats tend to follow their parents around so when the parents 
are among pine cones and picking them up the young rat also just because it's 
accompanying the adult rat will also be in the presence of pine cones also pick 
them up and it may learn the behaviour 
individually but just because it follows its rat it tends to do it its parent 
it tends to do it in the company of parents a slightly more sophisticated thing 
that the young rat might be learning is that pine cones are good they might be 
learning because their mothers spend a lot of time with pine cones there might 
be a kind of higher order conditioning going on whereas they associate pine 
cones with something good and therefore are attracted to pine cones themselves 
and to try to er try to get er t-, try them out and learn by themselves this 
what i'm getting at is that there are a number of simpler kinds of individual 
learning that might contribute to the young rats i-, to the way the young rat 
itself learns to interact with pine cones which do not necessarily involve the 
young rat actively learning its mother's actions or still less being taught by 
its mother as i have so often said in these lectures you have to discount 
simpler explanations before you can accept more complex 
explanations of what in this case the young black rat is learning so in the in 
in the later part of this lecture i'll be talking about how experimentally one 
might be able to distinguish between these rather simple explanations that what 
the young rat is doing individual learning but somehow enhanced by the presence 
of its mother or whether the young rat is really learning something by watching 
its mother that's the sort of question we have to address there are other 
there's another very well known example that i want to mention briefly and come 
back to at the end of this lecture which is the macaques in Japan macaques are 
a kind of monkey kind of rhesus monkey the macaques in Japan who learned to 
wash food they were these were macaques on an island colony which was 
provisioned artificially fed by people among the food they fed them were sweet 
potatoes these sweet potatoes tended to get sandy and dirty when they were 
lying on the ground and it was observed that 
certain individual macaques acquired the behaviour of dipping the potatoes in 
water or dropping them in water and then taking them out before eating them so 
they were nice and clean this behaviour was initially observed in a very small 
number of animals in fact in just one animal initially but it gradually spread 
through the colony more and more animals showed it and the question was was 
this social learning were the macaques learning from each other that it was a 
good idea to dip food in the water i'll come back to this question at the end 
of the lecture i won't stop with the rats diving for shellfish example because 
er i don't really have time but there are other examples where specialized 
behaviour such as diving to get shellfish from the bottom of a river are shown 
in just some colonies of animals and not others as i've tried to indicate these 
kind of observations while very fascinating do not themselves provide evidence 
of social learning i 
argued this in some detail in the case of the black rat but we can argue it 
more generally by saying that when you see a behavioural difference between 
different colonies of animals this might be due to learning social learning 
going on in one colony and not the other or it might be due to particular 
environmental conditions which are present in one group and not in another for 
example chimps who build nests might do so because the particular trees that 
they have exploited provide suitable material for making nests an environmental 
factor whereas chimps in other areas don't have such suitable trees perhaps if 
the chimps in other areas had suitable trees they too would learn to build 
nests individually er with the macaques watching washing food it could be that 
all the individual macaques are learning this by trial and error by 
instrumental conditioning without necessarily learning from each other and so 
on so we need some way to do controlled studies which 
manipulate what opportunities animals actually have to learn from each other 
and see whether they do in fact learn in the way that that is being speculated 
[cough] my first set of examples come from a-, and and i'm going to talk about 
some fairly classic experiments in this lecture but i would point out before i 
go on that there is a really excellent chapter on this subject in 
Shettleworth's book which is referred to in the reference list for this lecture 
Sara Shettleworth has a superb chapter on social learning it's called learning 
from others it's very up to date very thoughtful very comprehensive and i'm 
going to mention just a few of the examples that she mentions but if you 
seriously want to think about this area and it involves many complexities her 
chapter is a very good place to go anyway some of the best known work on social 
learning or putative social learning in rats in animals are about food 
preferences these are examples of learning the significance of 
stimuli learning what foods are good to eat and what foods are bad to eat and 
as i pointed out in connection with the young black rats many animals eat in a 
social situation they will eat in the company of other rats for ex-, rats will 
eat in the company of other rats young rats will often eat in the company of 
their adult companions so the question arises what do they learn for example 
it's very well known that young rats will tend to prefer the same foods that 
their parents eat and in a simple experimental demonstration of this Galef got 
some adult rats and trained them by prior training using the taste aversion 
learning procedure that i talked about last week that a certain kind of food A 
was good to eat but they also learned to avoid another kind of food B by means 
of taste aversion learning so the adults prior to the start of experiment had 
learned that food A was good to eat and food B was nasty and they were provided 
with both foods A and B and then young rats 
were put in with the parents and allowed to accompany them while eating and of 
course naturally the parents ate lots of food A but avoided food B and 
naturally the young rats did the same if they were subsequently offered a 
choice between A and B the young rats would also prefer A the question is what 
had they learned from the parents the first basic question is had they learned 
that food A is good had they learned that food B is bad the parents knew both 
these things but had the young rats learned both these things had the adults 
communicated if you like to the young rats both that food A is good and that 
food B is bad to test this Galef did the following simple experiment he tested 
the young rats alone with food A versus a novel food that they'd never had 
before if the young rats know that food A is good they should prefer it to the 
novel food and they did so the young rats had clearly learned a preference of 
food A now to test whether the young rats had learned to avoid food B he tested 
them again on their own with B versus C 
if they learned from their parents that food B was bad they should have avoided 
food B and preferred the novel food C but they didn't they showed no preference 
at all in other words although the adults rats knew that food B was bad they 
had not transmitted this knowledge to the young rats so the young rats have 
learned to behave to some extent in a similar way to the adults but they had 
not learned all the information that the adults knew they had learned that A is 
safe but they'd not learned that B is dangerous and this simple experiment 
makes a very general point that just because an animal behaves the same way as 
another doesn't mean it's acquired all the knowledge that the other has there 
may be simpler means by which the young rats come to behave in the same way as 
the adults for example it might simply be that food A has become very familiar 
to the young rats if you imagine what's happening food A is over here food B is 
over here the adults eat a lot of food A 
ignore food B the young rats follow the parents eat where they're eating so 
necessarily the young rats eat a lot of food A as well just because they're 
spending a lot of time next to the parents what that means is that food A will 
become very familiar to the young rats food B will remain very unfamiliar and 
that in itself would just be a reason for the young rats to prefer A because 
animals as i've mentioned earlier in these lectures prefer familiar foods to 
novel ones it might simply be familiarity it might be nothing more than that 
that simple experiments can't tell us which now the next experiment i'm going 
to talk about is one where adult rats learned by interacting with each other 
about what foods were good to eat so now we're talking about pairs of adult 
rats not young and adult rats and here the situation is much simplified because 
we don't have the situation where the young rats follow their parents around 
it's not a it's a much more straightforward situation 
the basic experimental observation which these studies are so which start from 
is this supposing you let one rat a demonstrator feed on a particular kind of 
food let's say it feeds on chocolate flavoured powder you then take that rat 
and put it in a cage with another rat who's never had chocolate powder before 
there is no chocolate powder present but the two rats will socially interact 
with each other and a lot of that social interaction involves sniffing licking 
grooming so that the observer rat the one whose never eaten chocolate powder 
before has a lot of opportunity to interact with the rat who's eaten chocolate 
and perhaps smell that food on the rat on on the demonstrator rat and you might 
well think that the observer rat would acquire a preference for chocolate as a 
result of this experience and indeed that's what happens observer rats who've 
been put in with a demonstrator who's previously fed on chocolate if they're 
then given a choice 
between chocolate and some other novel food they will prefer chocolate so 
clearly they're learning something from their encounter with the other adult 
but what exactly are they learning [cough] Galef did a series of experiments 
where he kind of simplified the information that the observer rats were getting 
by means of this simple apparatus rather than letting the rats interact with 
each other and giving the opportunity for all sorts of potential social 
transmission he simplified the situation so that he could present the the 
observer rat here with a very simplified version of the information from the 
demonstrator he did it by putting the observer rat in this kind of bucket 
affair in the wall of the bucket was just a tube in which another rat could be 
placed so it was not possible for this other rat to actually get in the bucket 
with the observer but the observer could int-, could be exposed to whatever 
sensory stimuli were coming from the demonstrator rat and the 
demonstrator rat had eaten a particular food and the question is under what 
circumstances would the observer rat learn to prefer the food that the 
demonstrator had eaten ignore these graphs 'cause i'm just going to summarize 
the results of these studies in a moment in fact this is this is number three 
on your handout so i've described the general er the general procedure what do 
observer rats learn when they have the opportunity to experience another food 
that er th-, another rat who has eaten a particular food and the observer rats 
were exposed to food X in a number of different ways and then on their own in 
the absence of the demonstrator they were given a p-, a choice between X and 
another food Y to see if they'd learned a preference and these were the kind of 
exposures that Galef could make in his bucket apparatus first of all you could 
have food X er smeared on the nose or mouth of the demonstrator rat and the 
demonstrator rat was just a normal awake rat so the observer could 
smell the food on the rat's face and also potentially maybe exchange some kind 
of information with the demonstrator the next ex-, the next condition was just 
the same except that the demonstrator rat was anaesthetized so it couldn't 
actively interact with the observer in any way but the food was still there on 
its nose and mouth in a third condition the rat who had the food on its nose 
and mouth was actually not alive the only difference was that in this case it 
was anaesthetized in this case it wasn't alive and you can see if you look at 
the result of the first three conditions that observers learned a preference 
for food X if they'd smelled it on the nose and mouth of a rat who was either 
awake or anaesthetized but not if they'd smelled it on the nose of the or mouth 
of a rat who was not alive it's important to realize people tend to assume that 
somehow the rats in condition three were avoiding a food 'cause it had been on 
the nose of the dead rat that's not the case at 
all they didn't avoid it they just treated it exactly the same as the other 
novel food they just treated it as it was completely unfamiliar they didn't 
avoid it all but in the first two cases they actively preferred it suggesting 
that for some reason they learned that this food is a good thing when they'd 
experienced it on the nose and mouth of an awake or anaesthetized rat so what 
was it that was crucial Galef tried two other conditions for example he said 
what would happen if you expose the observer to the food not on the nose and 
mouth but on the back of a live rat and in that case they didn't learn a 
preference either what happens if you just put it on a neutral base like a roll 
of tissue then they didn't observe then they didn't find a preference what 
Galef found in the end after these series of experiments that the only thing 
that was crucial there didn't have to be a rat there at all but that the 
observer rat had to experience the smell of the rat mixed with the smell 
of rat breath rat breath contains an unpleasant smelling chemical called carbon 
disulphide and what Galef found was that if he just exposed observer rats to 
the smell of food X plus the smell of carbon disulphide without any rat being 
there at all they would acquire just as much preference as if they'd smelled it 
on the face of a wak-, a waking rat what this implies is that there was no real 
social element to the learning in this case the observers had not learned to 
prefer the food X because of anything in the behaviour of the demonstrator rat 
but simply because if the demonstrator rat is alive that gives them the 
opportunity to smell it in the context of other rats' breath and this simple 
rule if you smell something in conjunction with a rat's breath it's good that's 
enough to produce this learned preference so this learning will normally take 
place in a social context other rats normally have to be around to generate the 
smell of rat breath unless an 
experimenter comes along and fools you by presenting you with carbon disulphide 
generally speaking you'll experience rat breath with a live rat but what this 
series of experiments shows is that the presence of the live rat is not 
necessary for the learning so this kind of experiment helps us to refine our 
understanding of what actually goes on as opposed to what might go on and it's 
it's very interesting to see how in real life a perfectly simple form of 
learning is enough to produce the desired effect that rats will produce foods 
which other live ra-, that rats will prefer foods that other live rats have 
eaten the next kind of stimulus significance learning that i want to talk about 
is learning about predators clearly and i don't have time to go into this 
animals have many ways of signalling the presence of predators to each other 
like alarm calls it's very common for animals to learn what's dangerous in 
social situations and the experiments i want to describe now are a very 
well known and elegant series of experiments by Susan Mineka about the way in 
which monkeys learn that snakes are dangerous in their natural environment 
these are monkeys these are macaques again who live in the Indian who live in 
India mostly in jungle or in areas surrounding towns where are there are a lot 
of snakes snakes are genuinely dangerous and wild-reared monkeys monkeys that 
have grown up in the wild are all frightened of snakes however if you take 
monkeys who've been reared in a lab have never lived in the normal environment 
and have never come into contact with snakes they are not afraid of snakes at 
all you can show them a picture of a snake or even a real snake they'll just 
treat it as a novel and slightly interesting object so it seems very plausible 
that monkeys in the in the le-, in the wild learn to fear snakes from other 
monkeys who've already acquired the fear and Mineka set up an experimental 
situation where observer monkeys could watch 
who were of course naive and didn't fear snakes initially as you'll see could 
watch a demonstrator who previously had learned fear of snakes for example a 
wild-caught monkey and the question is what would the observers learn from the 
demonstrator er to explain the procedure before i show you the data the 
observers were tested three times first of all a pretest when they were still 
naive and they'd never seen a demonstrator acting afraid of snakes 
a post-test immediately after they'd seen a demonstrator monkey acting afraid 
of snakes and then a follow-up three months later with no intervening training 
to see whether their f-, whether whatever they'd learned was persistent and the 
way the observers were tested was in a choice circus which was just a a round 
arena with four objects at the four corners one of which was a model snake and 
the other three were neutral objects and they simply measured how much time the 
observer monkey would spend near the snake if they were not frightened of 
snakes they'd spend about quarter of the time near the snake and a quarter of 
the time near the other objects if they were afraid of the snake they'd spend 
very little time near the snake and more much more time near the other objects 
so how much time they spend near the snake is one measure of fear the other 
measure of fear is that they used something called a Wisconsin test apparatus 
which is an apparatus simply where monkeys have to 
reach over a gap to get food and if you put a frightening stimulus in a glass 
box in the gap the monkeys will be reluctant to reach over it to get the food 
so in this test they put a snake either real or live in the glass box and look 
to see how slow the observers were to reach over the s-, over the snake to get 
a tempting bit of food and the slower they were and the more disturbed their 
behaviour the more frightened they were concluded to be of snakes so the 
question is how did the observers' behaviour change as a function of watching 
the demonstrators the graph is slightly complicated perhaps i'll just stick to 
the top one which shows the time spent near the snake in the choice circus and 
remember an animal that is er very frightened of a snake will spend very little 
time near the snake and much more time near other stimuli that are not snakes 
now this first graph here is not the behaviour of be observer but the behaviour 
of the model it's the behaviour of the 
monkey who really is frightened of snakes just to show you that if you really 
have a strong fear of snakes this is what you do in the choice appara-, in the 
choice circus mean time spent with each stimulus the stimuli are a real snake a 
toy snake a model snake and a neutral stimulus and you can see that the the 
monkey who really is frightened of snakes spends almost no time near either the 
real the toy or the model snake spends all its time near the neutral stimulus 
so that pattern indicates very strong fear of snakes what do the observers do 
okay here they are on the pretest when they're not afraid of snakes at all and 
as you can see they divide their time equally between the four stimuli they 
show no avoidance of snakes at all at the pretest but at the post-test when 
they've had an opportunity to watch an observer who is in the presence of a 
snake and acting frightened now they behave not as frightened as the model but 
very much more like the model 
they spend a lot of time near the neutral stimulus and very little time near 
the snakes so they have acquired fear of snakes just by watching another monkey 
and this fear is just as strong at the three month follow-up as it was 
immediately after so this is evidence that naive rhesus monkeys who are not 
afraid of snakes to start with can learn that snakes are dangerous just by 
watching another monkey they don't have to be bitten by a snake or attacked by 
a snake or anything they can just learn it by watching another monkey there's 
another very interesting feature of this learning which i want to mention 
because it it allows me to link this phenomenon with another effect that i 
mentioned in my lecture a week ago you remember in my lecture a week ago i 
mentioned that contrary to what Pavlov thought associative learning is 
selective some pairs of stimuli if you like are easier to associate than others 
so tastes are easier to associate with sickness visual stimuli are 
easier to associate with external pain in the case of rats there are many other 
examples now this observational learning of fear by monkeys turns out to be to 
have rather similar properties we've seen in the last experiment that monkeys 
can readily learn to be afraid of snakes by watching another monkey who shows 
fear of snakes but can they avoid can they learn to avoid other stimuli not 
snakes by watching an animal another monkey who's afraid of those other stimuli 
the problem is how can you induce a monkey to act fafraid of a harmless 
stimulus well for this purpose Mineka very cleverly just used a split screen 
video technique she made a video of a frightened monkey and at the bottom of 
the screen she showed either a picture of a snake or by splicing in another 
video a picture of a different stimulus which in her case was a bunch of 
flowers so the observers were in two groups one group of observers watched a 
video apparently showing a monkey who was frightened of 
flowers but not frightened of snakes that's F-L-plus S-N-minus meaning this 
model monkey was afraid of flowers but not afraid of snakes the other group S-N-
plus F-L-minus saw the opposite they saw a video of a monkey who acted 
frightened in the presence of the picture of a snake but not frightened in the 
presence of a picture of flowers and the question is would each of these groups 
learn to fear the stimulus that the model feared or as you might by now be 
expecting is there selectivity so that the observer monkeys can readily learn 
fear of snakes by watching a model but not readily learn fear of flowers and 
that's what happened over here is the group F-L-plus S-N-minus who watched a 
video of a monkey who was frightened of flowers but not frightened of snakes 
and you can see in terms of their reaction here again in this case slow 
responding indicates fear so i'm sorry this is the other way round from the 
last graph but this is the case where monkeys had to reach 
over a snake to get food so if they're slow at reaching that's a long latency 
that means they're afraid of snakes so what you're looking for here is is there 
a long latency in the presence of real or toy snakes the black and white bars 
so here we see the group who had watched a video of a monkey who was frightened 
of flowers but not snakes here's prewatching the video and post-watching the 
video and you can see that first of all there's no change in the attitudes of 
these monkeys to flowers they haven't learned to fear flowers at all by 
watching another monkey being afraid of flowers if anything they've increased 
their fear of snakes a little bit even though the monkey they watched in the 
video was not frightened of snakes they've still somehow er presumably by 
stimulus generalization linked the two the other group were completely 
different the group who've watched a monkey frightened of snakes but not 
frightened of flowers show a big increase in fear 
of snakes high latencies for snakes so this kind of associative learning 
observational learning also can show stimulus selectivity and i don't think you 
need me to tell you little stories to understand why that might be advantageous 
because snakes are a genuine predator therefore monkeys who can rapidly and 
efficiently learn fear of snakes from a small amount of exposure to a 
frightened conspecific are the ones who are likely to survive if they meet an 
actual snake so i just wanted to show you that example not only because it's a 
very elegant experiment which has also of course been used to cast light on er 
some of the fears both learned and unlearned that humans show because of course 
as you know snake fears are extremely common in humans even humans who've never 
met a snake in their life so Mineka's experiments have been used to discuss the 
possible biological basis of snake phobia but er i think th-, er but i also 
wanted to make the point that there is associative 
selectivity in that as well as in other cases the second kind of learning i 
want to discuss is the learning of new behaviours so far we've seen examples 
where animals learn something about the significance of stimuli in the food 
case they learned that food was okay but through a rather simple mechanism in 
the snake case they learned that snakes were dangerous ra-, in a rather 
specific way through watching other monkeys the next question is can animals 
learn complex new behaviours by watching others like the question can young 
black rats learn how to gnaw efficiently at a pine cone by watching their 
mothers well experiments on this topic have basically the following kind of 
design obviously you have one group let's call them the experimental observers 
who have the opportunity to watch a trained demonstrator who's already learned 
a particular action call it X and importantly gets rewarded for performing 
action X because clearly if observers are 
going to learn that performing this action leads to good things like getting 
food out of a pine cone presumably they need to see the demonstrator actually 
getting food for performing the action and in the test the observers are put on 
their own without the demonstrator and the question is how good are they at 
performing action X or perhaps how quick are they to learn it by themselves we 
need a control group to compare this group to various kinds of control groups 
are possible you might have control who watch a completely untrained 
demonstrator who doesn't perform action X at all you might watch nothing at all 
a better control is to watch trained demonstrators perform action X but not get 
any reinforcement for it that's much a much better balanced control 
and if you find that it's the expe-, the experimental observers perform X more 
or acquire action X by themselves faster than the controls they must have 
learned something from watching the demonstrators here's a simple example of 
such evidence this is an example that i took from Pierce's book er group E are 
the experimental observers who watch they're rats and they watched another rat 
performing action X which is pressing a lever and getting food there are two 
control groups control group one watch nothing at all they just watch an empty 
chamber so that's what i described as not a very good control group control 
group two experienced what i described as a good control procedure where they 
observed another rat pressing the lever so they saw the action but the action 
did not result in food finally all three groups of observers were put in the 
box on their own and allowed to press the lever for three days and in this case 
lever pressing did produce food so they 
had the opposite er the opportunity to learn for themselves that lever pressing 
produced food after watching these three possible demonstrations here's group E 
who had watched another rat pressing the lever and getting food and as you can 
see group E pressed the lever much more than the other con-, than either 
control group not only in the first session but they learned much faster as 
well so clearly group E had learned something from watching the other rat 
pressing the lever and getting food and what they'd learned was something 
different from what the controls grou-, from what the control groups might have 
gained from their experience it's interesting to notice by the way that the two 
control groups are very similar watching another rat pressing a lever and 
getting nothing really doesn't help you to learn any more than control one who 
watch an empty chamber so clearly watching the rat getting food and get wo-, 
pressing lever and getting food for it is quite crucial 
now once again something is being learned but we can raise the question what is 
being learned how sophisticated is the learning that's going on when the rats 
in the experiment i've just described are better at learning to press a lever 
for food when they've watched another rat press a lever for food what might be 
going on and Galef again who has been extreme-, one of the most influential 
researchers in this field set out some possible ways in which the experimental 
observers might have learned one very simple one which is er perhaps rather 
like a very simple explanation that i offered for the young black rats learning 
about pine cones is what Galef called local stimulus enhancement what he meant 
by that is this perhaps what the experimental observers learn when they watch 
the demonstrators perform action X and get food is that it's a good idea to be 
near that lever that because another rat is near the lever they get interested 
in the lever so when they're allowed to 
try it out on their own they spend more time near the lever and as a result 
they learn by the sort of processes i described in yesterday's lecture how to 
press the lever for themselves now you should immediately be able to see that 
that cannot explain the results of the experiments i've just shown you 'cause 
if that were what was going on you should have got just as much enhancement of 
learning in control group two who watched another rat press the lever and get 
nothing after all control group two had watched another rat pressing the lever 
so if it was just local enhancement getting attracted to that area of the box 
then control group two should show just as much lever pressing but they don't 
so clearly local enhancement as Galef called it while it's logically possible 
can't explain that example however what about the second possibility what Galef 
called blind imitation which means just copying what another animal does but 
not being aware of the 
consequences er you might say this is rather like kids who smoke because other 
kids smoke but don't actually care whether it makes them feel sick makes them 
ill later or whatever blind imitation without being aware of the consequences 
well once again this won't explain that example because otherwise again control 
group two who performed lever pressing without reinforcement should be likely 
to elicit blind imitation so that's not likely to apply here but one that 
certainly could be going on is what Galef called observational conditioning 
which is that what might be happening is not learning to associate the action 
of lever pressing with food but just learning to associate the lever with food 
perhaps the animals in the experimental group learn that every time they look 
at the lever because they're watching the rat food is delivered so they 
associate the sight of the lever with food whereas of course the other groups 
do not associate the sight of the lever with 
food it may not be learning the action of pressing the lever it may just be 
learning that that lever is a good thing it's associated with food and you all 
know what happens if you give that kind of Pavlovian training to rats if they 
learn to associate a particular object such as a lever with food what do they 
do they rush over to it they touch it they nose it and they're very likely to 
press it in the process get food for it and so this purely Pavlovian learning 
may set the scene for instrumental learning so this example could very well be 
what Galef called observational conditioning rather than what he called true 
observational learning I-E learning not just that the lever is good but you've 
got to press the lever to get food that's what Galef called true observational 
learning so how can one experimentally analyse which of these things is going 
on i've argued that not all of them could be going on in the rat case but 
certainly more than one of them could a very 
neat experiment to try and sort out these possibilities is this one by Palameta 
and Lefebvre who were French Canadians which explains their er slightly unusual 
names still are French Canadians in fact as far as i know er in their case this 
was an experiment with pigeons they had untrained pigeon observers who observed 
trained pigeons performing a particular action under a number of different 
conditions the action was this the model pigeons were in a box which contained 
two bits of paper a red paper and a black paper underneath the red paper there 
was a bowl of food underneath the black paper there was nothing it was easy to 
train model pigeons to perform a particular action piercing the red paper with 
their beak to get through it to get the food from underneath so the true 
observational learning group of observers watched precisely that they watched a 
demonstrator pigeon perform the action of piercing with its beak directed at 
the red paper not the black paper and they 
observed it getting food as a result so if true observational learning is 
possible it should be particularly good in that group the other groups were 
given experience that might be expected to produce either blind imitation or 
local enhancement and the question is how much improvement in the behaviour 
would they show first of all what did the groups have well the naive group were 
just a control group who never watched a model anyway so they're just the 
baseline the controls the blind imitation group watched a trained demonstrator 
pierce the red paper but not get any food so if the only learning that was 
going on was blind imitation just do the same thing as the demonstrator that 
group should learn just as much as the observational learning group did if 
those groups were equal it would suggest that the learning was blind imitation 
but if the observational learning group were better than the blind imitation 
group it would suggest that learning was more than just blind imitation 
that it involved learning this action leads to food the local enhancement group 
and this was quite cunning they saw the red paper and they saw the demonstrator 
put the head through the red paper and get food but because there was already a 
hole cut in the paper the demonstrator didn't actually have to perform the 
action of piercing so that group would see what should produce local 
enhancement or observational conditioning that this red paper is good 'cause 
the pigeon goes near it and it's associated with food so if that's the basis of 
the learning it should be just as good in these two groups so the argument is 
that th-, there was another pair of groups which were tested after a delay but 
i won't waste time with them so by looking so subsequently the observers were 
put in the apparatus and Palameta and Lefebvre measured how much tendency they 
showed how quickly they learned to perform the action of piercing the red paper 
themselves and the argument was if the true 
observational learning group was best that suggests that more is going on than 
just local enhancement or blind imitation but if either of these two groups is 
as good as the loc-, the observational learning group that suggests that the 
learning can be explained by local enhancement or blind imitation and does not 
require the full panoply if you like shouldn't use this word with namex here 
[laugh] he's testing out what language i use in my lecture you see i just 
realized that's a rather rare word anyway so if er pigeons who were learning 
the full experience if you like that this action is needed to get this food 
then only the observational learning group should be able to learn that and 
they should be best okay so what happened after all that forget the delayed 
groups we can just forget about them these are the four groups this is the true 
observational learning group who watched the model piercing the red paper and 
getting food 
this is their the latency of the observers to peck themselves over a series of 
trials these are like Thorndike's graphs showing the latency of responding 
getting faster and faster shorter and shorter as animals learn and you can see 
that that group really did learn fairly rapidly they took time on the first 
trial they weren't that wonderful they hadn't learned straight off but they 
fairly rapidly over a series of ten trials learned for themselves to perform 
the action none of the other groups was as good the group that was a little bit 
better than baseline was the local enhancement group suggesting that just local 
enhancement just learning that red paper is associated with food does produce 
some improvement so there is an element of local enhancement in this learning 
but much more of it is observational learning true observational learning the 
blind imitation group got nowhere suggesting that blind imitation just doesn't 
happen in this set-up and notice that these results are different 
from the results with the rats pressing levers there's no reason why the same 
learning mechanism should apply to pigeons learning to pierce a red paper for 
food and to rats learning to press a lever for food you have to do these 
experiments separately for each situation if you really want to know what's 
going on but the nice thing about this experiment is that it does demonstrate 
learning that an action leads to food and it rules out most of the simpler 
explanations except a very small element of local enhancement finally then er i 
don't have time to discuss all these cases but these are the conclusions i'd 
like to draw er we've seen from the lab studies that you can demonstrate these 
various types of observational learning but the underlying processes may not be 
obvious and they may often be much simpler than one would like to believe we're 
anthropomorphic we tend to believe that things go on in animals' heads similar 
to what would go on
in our heads when we're watching we think that if your mother shows you how to 
boil an egg you learn all about how to boil an egg and rats ought to be like 
that when they watch their mother chew-, chewing a pine cone they learn 
everything their mother knows that just isn't true animals' social learning is 
often a lot simpler than it would be in our case forget about the diving rats 
i'll leave that out because i haven't talked about it er i would like to talk 
finally very briefly about the famous primate examples you remember the case of 
chimps fishing for termites in a termite mound by putting in a twig and you 
remember the case of the macaques who washed food to what extent can we 
conclude that observational learning was responsible well in the case of 
termite fishing there's no doubt that young chimps intently watch their mothers 
when termite fishing and after watching their mothers they will often make a 
clumsy attempt to imitate they might get a twig they 
might wave it in the direction of the termite mound but just from watching 
their mothers they do not get skilled at it they do not learn from their 
mothers how to strip the twig they don't learn how to poke it in in a very 
stealthy way so that the termites will climb onto it without being ditu-, 
disturbed they don't learn the whole skill from watching their mothers after 
they've watched their mothers there is a long process of individual learning 
without necessarily another animal being around so they learn a little bit from 
observation but they don't learn the whole skill what about food washing well 
again there's been a lot of speculation about this example but i just want to 
show you i want to tell you two things about this finally the first is this 
graph here which comes from Shettleworth's book which is pretty simple this 
shows observation of this colony where you remember the behaviour of washing 
pot-, sweet potatoes spread through the colony this shows over a 
series of years the number of monkeys in the colony who were either o-, 
observed never to wash potatoes and as you can see there were quite a lot of 
those the number who were observed to wash potatoes starting with just one in 
nineteen-fifty-three and increasing to about fifteen by nineteen-fifty-eight 
and also a few who were in the process of learning to wash now what can we 
learn from a graph like this this line the number who actually wash potatoes 
goes up really quite slowly goes up from one to fifteen in five years so if 
this is observational learning it's very very slow the other thing is to look 
at the shape of this graph it goes up sort of steadily now if observational 
learning were really important you'd think that the more animals had learned it 
the more new animals would have the opportunity to learn 'cause if there's only 
one monkey who shows this behaviour it's going to be very rare for other 
animals to have the opportunity to watch her and 
very few will learn it was by the way a female who was a young female who was 
number one er but the more animals who were doing it you'd think the more 
opportunity other animals had to observe it happening so you'd think that this 
graph ought to go up more and more steeply whereas in fact it goes up rather 
slowly and Shettleworth argues that this kind of graph is much more suggestive 
of individual monkeys learning independently by personal by individual learning 
to do it rather than watching other animals one would like to believe that they 
do it by watching other animals but this kind of data is really not very 
compatible with that hypothesis it's much more compatible with individual 
learning individuals learning that oh this sweet potato tastes nicer if you put 
it in the water or maybe there was some local enhancement maybe they followed 
other monkeys to the water and thus had a better chance to learn themselves the 
other sad story about the monkeys on er Kashima 
Island is that it turned out later that you remember i've said that they were 
artificially provisioned they were fed by humans it turned out later that the 
humans who fed the monkeys were so charmed by this behaviour that they started 
training the monkeys to do it without the scientists knowing about it at first 
and in fact quite a lot of this increase could have been due to artificial 
training by humans rather than by the monkeys teaching each other now i i d-, i 
mean i'm ending these lectures on a note of scepticism i don't want you know to 
feel that there's nothing interesting to know about social learning in the wild 
there clearly is but the point i'm trying to make is that before jumping to the 
conclusion that some complex process is going on you really need to look a lot 
more closely at the experimental and behavioural evidence thank you