I. Definitions
It may seem that learning and memory would be an easy thing to talk about since they're such universal phenomena. But because they're so universal many people often use different words to refer to the same thing and sometimes the same words to refer to different things. That can lead to a wealth of confusion, so we're going to start by carefully defining our terms so that we're all on the same page, at least as much as possible, as we cover this chapter.
A. Learning
At its most general, we can define learning as a durable change in behavior or knowledge
due to experience.
This actually covers a lot of ground. So, a baby beginning to walk, a guy taking guitar lessons so
he play well enough to impress girls, a single mom taking real estate classes
to get her real estate license and switch careers, a kid figuring out how to
advance to the next level on a video game without resorting to cheat codes, a
grandfather regaining lost function in a hand after a stroke, all of these
reflect types of learning. These types of learning aren't all the same, don't all use the same psychological processes or require the same parts of the brain, but they're still learning according to our definition.
B. Memory
Using the same broad parameters we can define memory as the means by which past experience is
drawn on to guide or direct behavior or thoughts in the present. Like our previous definition of learning, this also covers a lot of ground. A guitar player effortlessly plays a song that he learned years ago, a student picking an answer on a multiple-choice test, a smell of a particular type of pie baking generating a vivid recollection of a childhood Christmas at your grandmother's house, an adult getting on a skateboard and riding it like he did when he was a kid 20 years ago. All of these are forms of memory, though they're all different, some involving conscious effortful recall of things you can put into words and other not. But they're still all forms of memory by our definition.
II. Some Types of Learning
We're going to focus on three types of learning. Each of them will induce a "durable change in behavior," but they differ in how they accomplish that change, what brain areas are involved, and even in what is learned and how that learning is applied.
A. Classical (Pavlovian)
Conditioning (involuntary responses)
Classical conditioning was first formally described by the Nobel Prize winning Russian digestive physiologist, Ivan Pavlov, in the 1920's hence its other interchangeably commonly used name, Pavlovian
conditioning. The simplest description of this type of learning is that
the organism (animal or person) learns a predictive relationship
between two external stimuli; the presentation of the first specific stimulus
predicts the imminent following presentation of the second specific stimulus. In a sense, the world is "happening" to the organism. The organism may have little to no control over its circumstances. But sometimes one thing always reliable predicts a second thing is about to occur. The first event then allows the organism to predict the
second event and prepare for it,
whether that second event is a good or bad thing, pleasant or unpleasant.
1. Acquisition
Pavlovian conditioning
starts with acquisition, the initial
learning of the predictive relationship between stimuli. Conditioning requires
that there first be an unconditioned stimulus (UCS), which is already known and established
through inborn instinct or prior learning to elicit an unconditioned
response (UCR) from the organism. In everyday English, a UCS is stimulus which already means something to the organism and already generates a response on its own, the UCR. An example of a UCS from Pavlov's experiments is food, which after the food is presented,
instinctively elicits salivation, a UCR, from a dog. If a stimulus event always occurs just before the dog gets
food, then that stimulus is a good predictor of the impending presentation of
the food (UCS). Such a predictive
stimulus, previously neutral and meaningless, is called a conditioned
stimulus (CS) and as the organism acquires
the predictive relationship the CS will elicit a conditioned response (CR) from the organism in preparation for the UCS
before the UCS is presented. In an example from Pavlov's work, a bell which was always rung before the food presentation was a CS. Prior to the pairings of the bell and the food, the dog might not have ever encountered the sound of a bell ringing in its life. But because the bell's ringing reliably predicted the food presentation, the sound of the bell eventually prompted the dog to salivate before the food was available.
This salivation that occurred to the bell before the food
arrived was the CR, a response in
preparation to predicted, expected opportunity to eat food.
2. Extinction
After acquisition,
should the CS be presented repeatedly alone, without the UCS following,
eventually the CR will cease to be elicited. This process is extinction. As an example, if the bell continued to ring for Pavlov's dog without food following, eventually the dog would stop salivating to the bell. This does not affect the UCS or the UCR. If food is presented by itself, the dog
will still salivate to the food, just not to the bell. However, extinction is not the same as
forgetting or erasing the prior CS-UCS association. What happens during Pavlovian extinction is that a new, yet
opposite, predictive relationship is learned; that the CS now predicts the
absence of the UCS. This new association now competes with the prior association. As evidence that the prior association
is not forgotten or erased consider this: If the same extinguished CS once
again is paired with the UCS, the association is re-learned even
faster than it was initially, indicating that the prior learning provides an advantage on the second attempt at learning. So, continuing with our Pavlov's dog example, if the bell was again paired with the food, the bell would elicit salivation after fewer CS-UCS pairings than it did the first time.
3. Spontaneous Recovery
Spontaneous recovery is further evidence that extinction is
not erasure or simply forgetting.
Extinction is typically incomplete after one bout of CS presentation without
the UCS following, even if after the first bout of extinction training the CR
was apparently completely eliminated.
It is common that if the CS is presented again during another training session,
the organism will perform the CR although with less vigor and effort and it
will soon extinguish again.
Sometimes several bouts of extinction, CS-alone, training are required
before spontaneous recovery completely ceases and extinction is complete. The reason is that each time the two associations, acquisition and extinction, compete and it may take several bouts of extinction training before the extinction learning can completely overwhelm the initial acquisition learning. Using our Pavlov's dog example, after the first day of extinction training, each successive day when the bell was presented alone there'd be some small amount salivation to the bell at the beginning of the session. It may
take several days of presenting the bell alone before all salivating to the
bell completely ceases.
4. Stimulus Generalization/Discrimination
In stimulus
generalization, stimuli with similar sensory qualities to the CS may elicit the CR, essentially by mistake. In our Pavlovian example, let's say that a dog has learned that a bell predicts the presentation of food. But one day somebody walks by the dog and their keys are jingling and the metallic sound causes the dog to salivate. That would be an incidence of generalization. But let's say that that person with the jingling keys starts to walk by often. Eventually, the dog will learn to salivate only to the bell and not to the keys jingling. That would be stimulus discrimination. We can make the discrimination more difficult. Let's say the experimenter now introduces a second bell with a slightly different sound, maybe a little bit higher or lower in pitch than the original bell. The dog may have a more difficult time learning to tell the difference compared to keys jingling. But with further training over time, the generalization to the second bell (as evidenced by salivation to the different bell) should decrease and the dog should be able to discriminate between the two bells' similar sounds.
5. Higher Order Conditioning
Higher order
conditioning is a phenomenon that Pavlov
used to explain how a series of conditioned responses to a succession of
stimuli may be chained together. Recall that our definition of a UCS stated
that it was already known and established through instinct or prior
learning. In higher order conditioning a
previously learned CS has become so well trained that it can serve as a UCS to
a new neutral stimulus that predicts its occurrence. In our example, consider what would happen if the
experimenter would start to project the image of a beach ball on the wall in
front of the dog every time before the bell which predicted food was rung. Because the image of the beach ball
reliably predicts the bell which reliably predicts the food, eventually the
bell will act as a UCS for the image of the beach ball which will act as a new
(higher order) CS. The image of the beach ball will now generate salivation as
well as a new (higher order) CR.
B. Operant (Skinnerian) Conditioning (voluntary
responses)
Operant conditioning (also referred to as operant learning) was first developed by B.F. Skinner in the 1930's. In operant learning the consequences
of a behavior come to guide and control the occurrence of that behavior. In contrast to classical conditioning, in operant conditioning the organism "happens" to the world. The
organism is constantly performing behaviors; some of those behaviors wonÕt have consequences and some of them will have consequences. When a behavior has appetitive (pleasant, appealing, rewarding) consequences the
likelihood of that behavior being performed again will increase. If the behavior has aversive (unpleasant, disturbing, punishing) consequences the likelihood of that behavior occurring again goes down. Thus, the outcomes of the behaviors come to control them. So, instead of the organism merely reacting to predictive cues about what is about to occur (the world "happens" to the organism) and trying to prepare a response, as in classical conditioning, in operant conditioning the organism first generates a response (the organism "happens" to the world) and the outcome of that response determines if the organism will perform that behavior again in the future. An example of this is the classic Skinnerian lever-press task. A rat in a small chamber will seemingly randomly explore the chamber. It may poke and prod and move and jiggle and climb and manipulate any and everything in the chamber with no outcome. But then, at one point,
the rat pushes down on a lever and a food pellet rolls out of a little hole in
the wall into a cup. The rat may
eat the food pellet and go on poking and prodding and exploring and after some
time push down the same lever again and again a food pellet rolls out of a
little hole and lands in a cup.
Soon the rat is no longer randomly exploring, but continuously pressing
the lever and eating until it is no longer hungry (or the experimenter runs out
of food pellets). The emitted
behavior (lever-press) becomes controlled by its consequences (presentation of
a food pellet).
1. Reinforcement vs. Omission/Punishment
One thing that is often a pet peeve of purist animal learning researchers is the misuse of the term "reinforcement" in popular usage.
Some of the more tightly-wound researchers sometimes shudder when they hear an angry parent say to a misbehaving child, "If you don't stop, I'm gonna apply some negative reinforcement!"
This is why; strictly speaking, reinforcement increases the likelihood of a
response. So, the use of the terms positive reinforcement and negative
reinforcement differ in the nature of the stimulus used as a reinforcer
(appetitive = positive; aversive = negative), but both would increase the
occurrence of a behavior. So if a
parent wanted to increase the frequency of a child cleaning their bedroom, the
parent might give the kid a few unexpected dollars whenever they saw the child
cleaning their room. Just like the
rat lever-pressing for food (also positive reinforcement) the consequences
would be expected to increase the
room-cleaning behavior. If the parent spanked the child and said, "You're gonna get another one if you don't start cleaning your room," that's negative reinforcement. An aversive stimulus is applied and that aversive stimulus is terminated when the behavior increases. When a
stimulus is applied to decrease
the occurrence of a behavior, as in a parent threatening to unpleasantly
discipline a misbehaving child to get them to stop their behavior, is punishment. So, if
a parent spanked a child to get them to stop a bad behavior (using swear words,
for instance) that is simple punishment.
However, all forms of punishment do not involve applying an aversive
stimulus. One form of punishment is reward omission, where an expected appetitive stimulus is removed or
withheld. An example of this would be a parent taking away a cherished toy as a
consequence for a child using swear words. Whether simple punishment or reward omission is used the aim
is the same, to decrease the occurrence of an unwanted behavior.
2. Schedules of Reinforcement
In regard to positive
reinforcement, different patterns of reinforcement produce very different
patterns of behavior. The rules or
patterns governing reinforcement are called schedules of reinforcement. These
schedules determine the manipulation of either the number of responses (the
ratio of responses) or the elapsed time
since the last rewarded response (the interval of responses) In the simplest cases, a fixed
ratio (FR) schedule dictates reinforcement
after a fixed number of responses have been made. For example, an FR 10 schedule would deliver reinforcement for every
10 responses while a FR 1 would deliver reinforcement fro each response (also
called continuous reinforcement). An FR schedule generates rapid and consistent responding. It is similar to human "by the piece" payment schedules in some jobs. A company that pays a fixed price for every finished product
produced by the worker, regardless of how long or short the time involved would
be paying the worker according to an FR schedule.
In contrast, a fixed interval (FI) schedule dictates reinforcement of the first response after a fixed time interval has passed since the previous reinforcement. For example, in a FI 30 schedule, regardless of the total number of responses, the first response made after 30 seconds has passed since the last rewarded response is reinforced. This tends to create a lack of responses during most of the 30 second interval with a flurry of responses near the end of the interval until the 30 second time period has elapsed and the first response during the new 30 second interval is reinforced and then the organism stops responding until near the end of the time interval in anticipation of a new interval beginning.
There are also variable ratio (VR) and variable interval (VI) schedules. These are similar to their fixed counterparts but variable aspect means that the organism has to deal with some uncertainty with each response. In a VR 5 schedule, for instance, on average every fifth response would be reinforced but the actual ratio constantly varies randomly after each reinforcement, sometimes the next response might be reinforced, sometimes the tenth, sometimes the third and sometimes the sixth. But on average, every fifth response in a VR 5 schedule would be reinforced. Similarly in a VI 20 schedule, on average the first response after 20 second had elapsed would be reinforced, but each time interval varies around that average.
Each type
of variable schedule does not generate as high a level of responding as their
fixed counterparts, but the organisms do acquire a kind of persistence due to
the constant uncertainty of the occurrence of their reinforcement. If reinforcement is suddenly stopped
with an FR or FI schedule, the organisms quickly cease responding. Often they
display a form of distress called frustrative nonreward because of the failed reward expectancy. Imagine how you'd feel if you worked at job all week and then the boss said you weren't going to get paid for your work. You'd be experiencing frustrative nonreward. However rats, for
instance, trained on variable schedules are far more persistent and display
little, if any, frustrative nonreward. Why? Because their training schedules
accustom them to dealing with uncertainty of reinforcement on each
response. Organisms trained on
variable schedules can be quite persistent in the face of nonreinforcement. As an example, humans at a casino
playing the slot machine are perfect examples of a VR schedule and you can see
how persistent those gamblers can be in their responding.
3. Comparison of processes Classical vs. Operant
All of the processes seen
in classical conditioning, acquisition, extinction, spontaneous recovery, stimulus
generalization and discrimination and higher order conditioning are also found
in operant conditioning. When outcomes occur after behaviors, there is
acquisition and then when those outcomes cease to occur there is extinction.
When two similar stimuli are present (similar looking levers to press, for
instance), then generalization and discrimination can occur. When successive outcomes can be strung
together with successive behaviors, higher order conditioning occurs. There is some debate as to why these
two different forms of learning exhibit such similarity. One view of the basis of this
similarity suggests that in operant learning the animalÕs own response serves
as a predictor (or cue) of the outcome, much like the CS predicts the UCS. From
this simple associative basis the rest of the similarities may be eventually
built.
C. Observational Learning
Observational learning, also sometimes
called social learning, is learning that occurs when an organism learns from
observing and duplicating behavior observed in other organisms. Unlike classical or operant conditioning, the organism is not directly involved. The world doesn't "happen" to the organism nor does the organism "happen" to the world. Only from observing events, actions and consequences occurring to and because of another organism does the organism learn.
For instance it has been demonstrated that rats can learn to find food
or avoid punishment by observing the actions and reactions of other rats in
various situations. Similar findings have been reported with birds and even ants. In humans the earliest descriptions were the work of Albert Bandura in the early 1960's, including his classic Bobo doll experiment.
Bandura was studying the social determinants of aggression in children. The Bobo doll was an inflatable clown
doll. Children, boys and girls, were exposed to an adult playing with variety
of toys, but eventually acting violently with Bobo doll, beating it, kicking it
and throwing it with no adverse consequences to the adult. Once exposed to that, children (both
boys and girls) were much more likely to imitate that aggressive behavior when
left alone in the room, as observers recorded their behavior behind a one-way
mirrored glass.
III. Memory Processes (Big
Picture)
The seemingly effortless recollection of facts depends
on several processes working together. While analogies to computers are often
used, the brain and the mind do not work exactly like electronic computers. For instance, there is evidence that learning and storing memories causes physical changes in the anatomy of the brain. Computers don't change the size or shape of their components as they store information. However, there are some similarities in the processes used. Those processes are encoding, storage
and retrieval. Though one thing
should be kept in mind. If for
instance you are taking a test, and cannot remember the answer to a question,
or misremember the information, at that point it is, practically speaking,
usually impossible to determine with certainty which of these processes may
have failed you.
A. Encoding
Encoding involves processing the sensory
qualities of the information to form a sort of code or representation of the
information to be stored. During
encoding the depth of processing can play
an important role in how well the information is stored and later
retrieved. In shallow
processing emphasizes only the physical
structure of the stimulus. For example, were the words written in capital
letters or in italics? In intermediate processing there is slightly deeper yet basic analysis of the
stimulus, such as phonemic processing of the word. An example of this would be sounding out a new word you read
and noticing that the word rhymed with another word you already knew the
meaning of. In deep
processing the analysis grows more complex
and detailed. For instance, you might determine the actual definition of the
word, its semantic meaning and understand how it might be used in a sentence. The better (deeper) the encoding, the
better the storage and the more cues there are for more efficient retrieval.
B. Storage
Storage involves retaining the
information. One thing to consider
is that once a memory is stored, even with adequate encoding, it is not always
stored. Human memory is moldable. It can be strengthened or weakened during a
process that occurs shortly after learning called consolidation. During the consolidation phase the durability of
memory storage is either reduced or increased. For instance, you might want to remember specifically where
you put your car keys for several hours or may be even overnight but not for several days, let alone years from now. But should you get married, you might want to remember the names of who was in your wedding party for the several hours of the ceremony as well as weeks or years later. So, some memories get preferred storage while others don't. Consolidation is not a foolproof
process. It can also be disrupted by such factors as distractions and
competition between memories for the mental and biological resources necessary
for storage.
Two types of interference are constantly at work while we try to learn and remember. The effect of these interference types is largely inescapable because we can not completely stop learning and storing information. The processes of learning and memory are very closely tied to simply processing sensory information, so if we are conscious and aware, to some degree (greater or lesser) we are always using the mechanisms used to learn and form memory. One type of interference is proactive interference, where old information in the process of being consolidated interferes with the storage of new information. Conversely in retroactive interference the consolidation of the new information interferes with the storage of the old information. So every bit of new information we try to store is constantly struggling with older and newer information for the resources necessary for proper storage in a constant two-sided struggle.
This two-sided struggle is the basis for a phenomenon called the serial-position effect. If a list of words is simply read or heard by a subject, no special instructions provided, and later experimenters ask for the subject to recall as many of the words as possible, what researchers report is that the first word and the last word are the most commonly and easily recalled, with words in the middle of the list most poorly recalled. The enhanced recall of the first word is called the primacy effect and the enhanced recall of the last word in the list is called the recency effect. The enhanced recall is the result of the words only having what is essentially a one-sided struggle with interference while the words in the middle of the list have poorer recall because their storage was subject to both proactive and retroactive interference from the other words around them in the list.
One
strategy to minimize interference is to allow for multiple breaks while trying
to learn new information. Because of that, cramming for an exam is not the best
use of time, effort and energy. We can compare massed practice (essentially, cramming, one big bout of attempting
information processing and storage) with distributed practice (the same total amount of time spent attempting the
processing and storing of information). With the same total amount if time
spent trying to learn, distributed practice yields better test recall performance
than massed practice. Presumably this is due, at least in part, to the reduced
total proactive and retroactive interference experienced with distributed
practice.
C. Retrieval
In retrieval, the memory must be
accessed and brought back up to mind. If the information was well-encoded with ample
connections made during deep processing to information that is already well
learned and stored and the information was stored with a minimum of
interference, then the retrieval process has an optimal opportunity to access
and call up the information.
Retrieval can be compromised by emotional distress (fear, anxiety,
depression). These emotional states can produce a temporary retrieval
failure. However, should they
subside, often retrieval can suddenly occur. I'm sure almost everyone reading this text has experienced blanking out on the answer to a test question and as soon as they walk out of the test environment, the correct answer they were blanking on pops into their head.
Now that we've covered these basic processes let's see how they and other processes work together to take a sensory event and make it into a long term memory. The first stage is sensory memory. Sensory memory is very high capacity and highly accurate, but it is very, very short lived, along the lines of one second. In practice, it is constantly being refreshed by the sensory stimuli being processed via the sensory systems of the brain. In the absence of that constant updating it rapidly decays. The process of attention focuses mental resources on one element out of sensory memory and brings it into working memory (also called short-term memory). Working memory starts to decay noticeably after 30 seconds and lasts maximally about 3-5 minutes. It also has limited capacity, able to store 7 +/-2 "elements" or "chunks" of information. The life of an item in working memory can be extended through what is called maintenance rehearsal, basically repeating the information over and over. As an example, if someone gave you their phone number and you couldn't write it down, the seven digits would correspond to the seven information elements. Repeating the number over and over while you looked for a pen and piece of paper would be maintenance rehearsal. Now, the transfer into long-term memory (which can last for minutes to potentially for a lifetime) from short-term memory can be made easier by elaborative rehearsal, where the information is more deeply processed and often organized into you pre-existing long-term memories. To continue our example, let's say phone number was 339-1233. For me personally, I might organize it by the first three digits being the mathematically related (3 x 3 = 9) and because I've been a Dallas Cowboys fan since I was a small kid, I might organize the last four digits by the jersey numbers of two historic Dallas players, in this case Roger Staubach (# 12) and Tony Dorsett (#33). That would be a type of elaborative rehearsal. Therefore in retrieval, I would have those bits of well-known information (my elementary school multiplication tables and my childhood sports heroes) as retrieval cues to help me call up that phone number.
Now the durability of the information in long-term memory is determined by the consolidation process. WeÕve seen how distractions or interference can impair consolidation, but consolidation can be improved by two things, repetition and perceived importance or interest. Simply repeating the same information, studying it over and over, will help consolidation and improve recall performance. But more potent that simple repetition is importance or arousal. The more engaging or interesting or important the information is judged by the subject, the better the recall performance later. As we'll see later, there are very important biological mechanisms that underlie the facilitating effect of interest or arousal on consolidation.
IV. Forgetting
Forgetting is a complex and not entirely well
understood process. One theory is
that the memory simply decays, that the information simply fades, possibly erased or "overwritten" in the brain. Much like the memory for what socks you wore three years ago last Tuesday. The interpretation there is that that memory was not deemed critical and was weakly consolidated and allowed to fade. Another theory is based on interference with retrieval. This holds that the retrieval of a given memory is impaired by other, especially similar, memories. The information may be there waiting for a strong reminder or retrieval cue. But when retrieved it may also be contaminated or blurred by those similar memories which might have been interfering with its recall.
The take-home message is that human memory, while generally very good, long-lasting and accurate is not perfect or infallible. Despite all the processes and factors which can make memory better, we can still sometimes make honest mistakes based on our memories under the best of circumstances.
V. Learning & Memory
(The Biology)
The brain mechanisms of learning and memory rely a
great deal on the forebrain systems we covered in the last chapter which dealt
with planning, foresight, motivation, emotion and monitoring our internal
physiological state. Our complexity of learning and memory capacities and
abilities reflect to the range of complexity of our cerebral cortex and our
forebrain systems in general, all the way down to the regulation of
long-lasting microscopic changes to our synapses. The changes to the strength
of communication at the synapses are called synaptic plasticity,
meaning simply that the strength of the synaptic connections is changeable but
stable. Learning and memory is
just one type plasticity. The same
molecular mechanisms seem to also be involved in other long lasting changes in
the brain such as growth and development, addiction, and recovery from an
injury such as a stroke or concussion.
What makes some examples of synaptic plasticity involved in memory is
determined by the brain systems where that plasticity occurs. Plasticity in
brain areas involved in memory then serves a memory function. In other brain
areas, that plasticity would serve the function that those brain areas are
involved in.
A. The Limbic System
The limbic system, particularly the hippocampus, amygdala and the cortex of the medial temporal lobe (as well as the prefrontal cortex which is technically not part of the limbic system but receives numerous projections from the structures of the limbic system) are extremely important for what we would normally informally refer to as our "memory." The hippocampus (and to some degree the
cortex of the medial temporal lobe) is responsible for moving information from
short-term memory to long-term memory.
Neither form of memory depends on the hippocampus for its existence, but
it is necessary for information to pass from short-term to long-term.
The prototypical case is that of the patient, H.M., who as a teenager in the 1950's had experimental (for that time) surgery to bilaterally remove the hippocampus from each temporal lobe to treat an extremely severe case of temporal lobe epilepsy which was resistant to medication and threatened to kill him. After surgery, it was found that his old long-term memories were intact and his working memory was intact but he could not form new long-term memories. A stranger could introduce himself to H..M., hold a brief conversation with him then leave the room and come back three to five minutes later and say hello and to H.M. would have no idea who that person was nor have any memory of ever having had a conversation with him. H.M., today is an old man and while he has grasped that something is wrong with him, he basically has lived in a day-to-day world where Dwight Eisenhower is the President of the United States, no man has ever ridden a rocket to the moon, Frank Sinatra just had a hit song last year, and the US flag has 48 stars.
The amygdala is involved in attaching emotional significance to the information the hippocampus transfers into long-term memory. The amydgala's role can be seen to be the biological basis of the role of importance or arousal in facilitating the storage of memory. It is not absolutely necessary to form a long-term memory. But because of its connections with the hippocampus it can make the hippocampus' job easier. However, without the amygdala the information is just cold fact without emotional impact. Someone with amygdala damage might see a particularly bloody and gruesome photograph, for example, and while they might be able to remember the specific details of the photograph later, they wouldn't show any signs of emotional reaction while recalling the information.
B. The Frontal Cortex/Basal Ganglia Circuit
The frontal cortex is involved in planning and
foresight, even to the point of selection actions and behaviors. The basal
ganglia, as we covered in the last chapter, is involved in guiding and
directing voluntary movements. There are connections from the basal ganglia to
the thalamus that then the thalamus relays to the prefrontal cortex. Every time
a complex motor action is selected by the prefrontal cortex and successfully
guided by the basal ganglia, particularly the caudate nucleus and putamen, the
loop via the thalamus back to the prefrontal cortex helps to strengthen the
activity of those circuits. This process is believed to be the mechanism by
which practiced movements become habits, able to be performed automatically and
without conscious effort. So, whether it's playing a guitar, riding a bike, playing a piano, dancing a folk dance, the intentional conscious effort first directed by prefrontal cortex eventually becomes automatic.
It is interesting to note that this system is completely independent of the limbic forms of learning and memory. One could teach somebody with bilateral hippocampal damage with the same sort of memory impairment as H.M. to play the piano. That person could practice every day. TheyÕd have no memory of ever having practiced If you asked them is they could play the piano, they'd deny it. If you begged them to try, they'd be able to play but they would be at a loss to explain how or why. Their frontal cortex/basal ganglia circuit would have learned independently and separately from the damaged limbic system circuit.
C. Long-Term Potentiation
Long-term potentiation (LTP) is a biochemical mechanism of synaptic
plasticity that is important fro memory. LTP occurs at glutamate synapses and
requires the activation of the normally silent NMDA receptor to be induced.
The biochemical changes caused by NMDA receptor activation then cause
modification of the other ligand-gated glutamate receptor, the AMPA, to
maintain expression of LTP. Under usual conditions of activity at a glutamate
synapse, the AMPA receptor functions but the NMDA receptor is blocked by a
positively-charged Mg++ ion that is wedged in the pore. The inside
of the dendrite needs to be made positive by AMPA stimulation for an unusually
long time for the Mg++ block to be removed. So, functionally, only
an unusually great amount of activity can trip the NMDA receptor. We can think of this a case where only
an extremely important signal can cause the plastic changes of LTP, which can
serve memory storage. Once the
NMDA receptor has been activated, Ca++, not Na+, enters
the dendrite via the NMDA receptor.
Just as Ca++ was absolutely critical for presynaptic
transmitter release, it is also absolutely critical postsynaptically for the
changes underlying LTP. The Ca++
triggers biochemical changes, such as enzyme activation, that modify the AMPA
receptor so that the same normal amount of glutamate as before causes a much
larger EPSP on the receiving neuron. The strength of that synapse has been
increased. The receiving neuron,
as a whole, can now can respond much more easily to signals from that synapse.
Many researchers believe this synaptic long-lasting plastic change is the basis
of memory storage.
D. Multiple Memory Systems
We have already seen that all memory is not the same. Short-term
and long-term memory differ. The type of memory supported by the limbic system
is different from the type of memory supported by the basal ganglia. Research
has divided memory in several ways.
WeÕll discuss some of them here.
One general category of human memory is called declarative memory, which stands for fact-based memory that can be expressed (declared) in words. There are two subtypes of declarative memory. One is semantic memory, which is general knowledge of the world. It is non-autobiographical and we typically can't recall when or were we acquired the information. Examples would be things like Columbus set sail in 1492, the first meal of the day is called breakfast, a red traffic light means stop and a "dozen" means 12 items. All facts that can be declared verbally, but just non-personal knowledge of the world. The other type of declarative memory is episodic memory. This is what we mean when we usually talk informally about memory. Episodic memory is our personal autobiographical recollections. The story of our lives. While it is semantic memory to describe what somebody might reasonably eat for breakfast or wear to school, it is episodic memory when we relate what WE had for breakfast or what WE wore to school. The limbic system is important for declarative memory but is especially critical for. episodic memory. However the limbic system is not critical for all forms of declarative memory. For example, temporal lobe-limbic system damage does not impair working memory, but the prefrontal cortex does play a role in maintaining working memory.
Another category of memory is procedural memory, which is memory that does not or cannot be verbally transmitted or declared, it must be expressed as an action. Habits and classical or operant conditioning are example of this type of memory. While the basal ganglia is critical for habit formation, it does not control all procedural learning. For example, one type of classically conditioned eye-blink reflex to a tone which predicts an air puff to the eye is controlled by a circuit completely within the brainstem and cerebellum.
E. Memory Modulation
We've already discussed that the amygdala adds emotional content to the information that the hippocampus transfers to long-term memory and facilitates the transfer. WeÕve discussed that perceived importance, interest or emotional arousal makes memory formation and consolidation easier. These are all examples of memory modulation. Memory modulation occurs largely as the
consequence of a physiological and hormonal stress response to an event. The
hormonal responses and their consequences serve to mobilize the body to deal
with a crisis. But they also serve a second duty, to facilitate the storage of
information during that crisis period. We can agree that should we survive a
stressful crisis period, the information gained during that time would likely
be valuable and should have its storage enhanced in case that situation should
arise again.
The amygdala is involved in triggering the physiological response to stress but also in charge of orchestrating the memory-altering roles of the stress response as well. The key components of the stress response are, in humans, the steroid stress hormone cortisol, the catecholamine stress hormone epinephrine (also known as adrenaline) and the rise in blood glucose triggered by epinephrine. Cortisol can cross the blood-brain barrier to affect the amygdala and other brain regions. Epinephrine cannot cross the blood brain barrier. However, receptors for epinephrine on sensory peripheral nerve endings cause neuronal firing which eventually cause the locus ceruleus to release norepinephrine in the amygdala and throughout the brain. Epinephrine also causes a rise in blood glucose, to fuel the body's metabolic needs, but that glucose crosses the blood brain barrier to facilitate the brainÕs ability to store information as well.
This
memory modulation that arises due to stress is complex. There is evidence that
too little emotional arousal as well as too much emotional arousal results in
poorer memory retention or consolidation compared to a moderate level. This is
referred to as an inverted-U response curve; too little and too much low
performance, some mid-level high performance. Basically, when emotional arousal is low (as in a case of
boredom or disinterest) this basic memory modulation system is non-functional.
As emotional arousal increases, memory performance increases. But after a peak,
additional emotional arousal ( and release of stress hormones) starts to yield
less than peak performance. At
high levels of arousal, performance can be as poor as with little to no
arousal. The reasoning there is
that at higher levels even storage of non-important information is also
enhanced and, as with retroactive and proactive interference, resources for
storing information become overtaxed and overall interference is increased,
leading to poor memory performance.