|
Outline |
|
Introduction to Memory |
|
Part I: Definitions of Learning & Memory |
|
Part II: Types of Memory |
The brain is the most complex structure known. Neither philosophers nor scientists can yet explain how a mushy 3-pound collection of cells and fluids gives rise to consciousness, learning, and locomotion. What we do know is that brains allow us to adapt to our environment on a minute-to-minute basis, rather than the generation-to-generation pace of genetic mutation and natural selection. When natural selection evolved brains, it was something like a mounting a jet engine on a Conestoga wagon.
In everyday conversation, we use the terms memory, learning, and forgetting in very casual ways. From a scientific perspective, it is helpful to define these terms more carefully.
Most generally, we consider learning to be an adaptive change in
behavior caused by experience. In other words, learning is an observed,
external change in activity that is important to the welfare of the
organism. Memory
refers to the internal storage and recall of previously learned behaviors.
(Forgetting is the loss of
storage or recall.)
Memory must occur in some physical structure. In computers, most of us are
familiar with the storage of information as binary magnetic charges on floppy
and hard disks, or of tiny optical pits on CDs and DVDs. Biological memory must
also involve some structural changes—but the biological nature of the brain
requires that the changes be in the dynamics of the microcircuitry, rather than
as physical bits of information. That is, memory (storage/retrieval) must
result from specific biochemical and electrical processes. We'll elaborate on
this in the classroom lecture.
|
|
In mammals, memory appears to primarily involve the cerebral cortex, that wrinkled-up outer covering of the brain. Most of it is neocortex (i.e., more recent in evolutionary terms), although, for reasons of experimental convenience, a lot of research has focused on the archicortex, or hippocampus, a small "sea horse" shape structure that connects to the medial temporal lobe. During the course of the evolution of mammals, the neocortex expanded by over a factor of 1000 in size; in humans, almost 80% of the brain is neocortex (of which about 40% is frontal, and 20% each temporal, parietal, and occipital). It is not surprising, therefore, to find that research points to the cortex as the home of memory and intellect, and the basis for complex learning.
Importantly, although we have a lot of cortex, about 2/3 of it is white
matter—that is, axonal wiring insulated by (white) glial myelin sheaths. Think
about connecting N cells to one another. It would take N*N axons. That is, the number of connections grows as
the square of the number of cells. There are about 20 billion neurons in the
human, connected by about 150,000 km of axonal wiring. That’s half the distance
from the earth to the moon! Clearly there are anatomical/vascular limits to
brain growth—nature had to evolve very efficient topology to handle the job.
(If you enjoy comparative facts, click
here for a great summary by Eric Chudler from the
|
MEMORY QUESTION 1 |
Preface
As you are aware, researchers like to classify things, even if they overlap considerably (for instance, arbitrarily declaring a glucose value above 125 mg/dl to be "diabetes"). The same brain areas may be involved simultaneously in several types of learning over several time scales. So it’s hard to isolate the subsystems involved in specific types of memory behavior. Are there really different physical brain networks for different types of memories? And are short- and long-term memory served by different systems, or are they part of a continuum?
Unfortunately, present brain recording technology is limited to two spatiotemporal extremes: recording directly with micrometer spatial and millisecond temporal resolution from a small number of neurons (i.e., not much of a functional network), or functional anatomical studies using PET (reflecting blood flow) and fMRI (reflecting hemoglobin oxygenation) which use relative metabolic changes or asymmetries over distances of centimeters (millions of cells) and tens of seconds. Neuroscience is still awaiting a technology that can track the simultaneous activity, at 10-millisecond resolution, of identified neuronal clusters across centimeters of brain. Even more challenging for the next Nobel prize winner: the new technology should be able to pick out clusters of cells which overlap spatially, while discriminating excitatory and inhibitory spiking activity within each cluster.
In the classroom lecture, I’ll describe some recent pharmacological experiments that attempt to dissociate different kinds of memory (it seems that there may in fact be distinct, overlapping systems). For now (and for testing purposes), we’ll stick with a fairly traditional classification of memory…
Functional Classification of Memory
Although we're not yet sure exactly how the biological microcircuitry of the brain gives rise to memory, researchers infer from studies that there are functionally two classes of information management. Declarative memory is the storage and recall of information available to the conscious mind, which can therefore be expressed (declared) using language. Declarative memory is also referred to as explicit memory, because it can be explicitly recalled; it seems to be localized solely in the neocortex. Declarative memory is subdivided into 2 types: semantic memory (facts and knowledge) and episodic memory (memory for specific events and experiences). The information you (hopefully :) remember from this tutorial involves semantic declarative memory in your neocortex.
On the other hand, procedural memory is the storage and subconscious retrieval of skills and associations of which we are not consciously aware. Procedural memory is also referred to as implicit memory, because its presence can only be inferred from observing behavior or performance. This usually involves complex motor skills, like walking, driving, skiing, and typing. In fact, you may even mess up your motoric performance if you start consciously thinking about the particulars of what you are doing. Procedural memory is active in cortical as well as subcortical (e.g., basal ganglia, cerebellum) brain structures.
Temporal Subclassification of Memory
Declarative memory can be subclassified by the time course of what is remembered. (One could also consider short- versus long-term procedural memory, but there is little reported research in this difficult-to-study area.)
Lasting just a few seconds, immediate (or iconic) memory involves the rapid decay of a sensory experience. Studies show that a series of sensory stimuli (e.g., photographs) can be shown at rates of up to 20 per second, yet be correctly picked several seconds later out from among other photos never seen before. But what is the purpose of immediate memory? Importantly, it constitutes the universe of activation patterns that could potentially be remembered. Fortunately, most immediate memory fades rapidly, leaving what is relevant to the animal; otherwise, the brain would be overwhelmed with useless information. However, some of it may be held in working memory by activation of executive circuits (below), and some of that may go on to short-term or long-term storage.
Moreover, the decay of immediate memory seems to be more than just "leftover" brain activity. Through an overlapping of a series of decaying experiences, the brain is provided with an internal clock pointing indelibly towards the future; it provides the linear vector that defines the flow of time of our existence. As such, it may be the neuronal basis for our awareness of the environment. Immediate memory has conscious components (declarative memory—what you think you are presently experiencing), but on a second-to-second basis is generally a subconscious process (procedural). However, even when subconscious, it can be useful to direct your attention to conscious events. For example, if you are listening but not paying attention to a lecturer, it may not be until several seconds after the word "exam" is spoken before you suddenly "hear" the sentence and perk up.
Short-term memory (STM) involves the retention and recall of events for a period of seconds to minutes. STM has a limited storage capacity—you may have heard of studies that find STM can hold “7-plus-or-minus-2 chunks” of information).
In the mini-mental state exam, you test this by asking the patient to remember, then recall 3 items. (You’ll perform the MMSE with a partner in class, so please look it over in advance.) Short-term memories can be temporarily sustained (potentially affected by the executive function of working memory) and/or go on to be permanently stored (long-term memory, LTM). We’ll discuss in class whether the short-term memory systems are needed for long-term memory to occur.
Electrophysiological studies in monkeys, and recent studies PET and MRI studies of humans indicate that working memory is a dynamic interplay of activation between cortical regions of information storage (location depending upon the sensory modality) with two frontal brain regions generating rehearsal and executive processes. For example, when asked to remember visual or verbal symbols, the left posterior parietal cortex is consistently activated. As the person or animal is required to retain the symbols for longer periods of time, Broca's area (lateral frontal cortex) becomes progressively more active (rehearsal). Broca's area may be a buffer, or temporary circuit that can mirror signals back to the posterior parietal region, in the absence of further sensory input. As expected, as Broca's area becomes more active, so does the parietal region, supporting the notion of a feedback loop. Of course, to prevent infinite looping (seizures), some meta-processes must control how long the rehearsal loop is sustained, and how attention of the person or animal should be directed. More on this during the classroom lecture.
Ultimately, information (or skills) with a lot of relevance and/or rehearsal (or practice) may become permanently available (the so-called engram). Most researchers believe that such long-term memory is possible only if there is some modification of the synapses between neurons involved in the experience. Although early experiments showed that connections between cells could be strengthened following prolonged, high-frequency stimulation (long-term potentiation, LTP), only in the past few years have neuroscientist begun to isolate synaptic responses to true physiological stimulation patterns. Here at UNR, we are collaborating with a leading group in the field, incorporating the findings into large-scale brain-like computer models.
It is important to realize that changes in synaptic responsiveness that result from learning do not, in themselves, constitute memory. It's not like computer memory, where the actual letters, numbers, or sounds are directly represented by binary coding on the hard drive or CD-ROM. The synaptic changes control the dynamics of the electrical activity passing back and forth among excitatory and inhibitory neurons (see movie, below). The more accurately the patterns of neuronal firing are regenerated, the more similar the memory will be the original experience.
|
|
|
Click to play a simulated activation of a stored memory. The left and right cell groupings represent two interconnected cortical regions. Pyramidal cells are excitatory (different colors in different layers), and inhibitory cells are red. Prior to a training exercise, there was only rare, random spiking of a few neurons. The movie above shows the activation pattern after training (i.e., after permanent synaptic changes). The activation pattern is reproducible, and terminates when inhibitory activity exceeds excitatory activity. (P. Goodman, UNR) |
|
MEMORY QUESTION 2 |
Many researchers believe that conversion of short-term declarative memory into long-term declarative memory requires the hippocampus (although some of us remain unconvinced). Their belief is mainly based on "natural experiments" involving the study of patients undergoing surgical procedures for intractable epilepsy clearly show that short-term memory does not necessarily lead to long-term memory (e.g., the famous case of H.M. described on page 556 of Purves et al, Neuroscience textbook). However, all of these surgeries removed more than hippocampus, and the patients were not neurologically normal to begin with.
The term "forgetting" can mean that information was either never stored (if pathological, we call it anterograde amnesia), or that it was stored but cannot be retrieved (if pathological, we call it retrograde amnesia). As mentioned above, most of the incoming environmental information and resultant thought content are (fortunately) not retained beyond immediate or working memory.
Forgetting short-term information reflects the desirable ("healthy" forgetting) or pathological interaction among the three components working memory: experiential cortex (usually the relevant sensory association cortex) and two frontal regions subserving the rehearsal and executive processes.
Forgetting long-term information usually means wide-spread damage to a cortical region, and is generally seen after head injury or during the course of dementia (to be discussed in a later neuroscience lecture on dementia).
|
MEMORY QUESTION 3 |
OK, you're ready for the MEMORY QUIZ...Good luck TAKE THE MEMORY QUIZ