This first in a series of posts about the biology of learning and memory has been written plainly, with much care taken to ensure its general-audience accessibility.
TL;DR: neurons communicate with other neurons by releasing chemicals called neurotransmitters onto them; these chemicals bind to receptors on the neighboring neurons. When a neuron repeatedly “fires” on another neuron, this triggers the latter to produce additional receptors and can even stimulate physical growth toward the former; thus, the latter becomes especially sensitive to the former and more likely to fire when the former does. This increased association, called LTP, can last a while, but it can also be extinguished.
In your brain, as in my brain, as in a sea slug's brain, connections between neurons get strengthened (or weakened) over time in response to increases (or decreases) in their activity. This strengthening of connections through repeated use is called Long Term Potentiation (LTP), a process which forms the neurochemical foundations of learning and memory. And while it is not too difficult to begin to understand, though it does require some background in biology to really grok the significance of the thing.
To get a handle on LTP, you have first to understand a bit about “electrochemical gradients.” You probably remember from chemistry that atoms can have a positive or negative charge when the number of protons (+) and electrons (-) are not equal. This is all “electrochemical” means here: these atoms (or molecules) with a net charge (+/-) are called ions and basically serve as chemical carriers of electricity. Everything we do is controlled by these electrical signals running through our bodies. Your thoughts, your emotions, your heartbeat, everything!
Now, you may rightly suspect that negatively charged molecules will attract positively charged molecules (and repel other negatively charged molecules). E.g., table salt is made up of two ions–sodium (Na+) and chloride (Cl-) stuck together like a opposing poles of a magnet. But what if we could somehow work against this attractive tendency and separate the two charges? Imagine we put all the (+) charges on one side of a wall and forced all the (-) charges to stay on the other side; this is analogous to holding two bar magnets close together, (+) to (-), but not letting them touch.
Just like damming up a river, this artificial impediment to the natural order (equal mixture of +/- charges) sets up a “gradient”. Like a ball being held on a steeply graded incline, the (+) want to flow “down” toward the (-), and likewise the (-) toward the (+), resulting in some serious “potential” to do work. When given the opportunity (say, by opening a door), the positive and negative charges will rush to the other side until the number of (+) and (-) charges are distributed equally throughout. This tendency to equalize the charges in space is fundamental principle of electrochemical gradients.
OK, now that all of the charges are spread out evenly on both sides of our wall, imagine for a moment that we take all of these charged molecules and put them all on one side of the wall, leaving nothing on the other side, and shut the door. There is no electrochemical gradient now; the net charges on either side of our wall should be zero (assuming the +/- charges exactly cancel each other out). But there is still a gradient! This time, we have created a concentration gradient; the natural state of the system is to be evenly spread out in space on either side of the wall, but we have concentrated all of our molecules in only half of the available space. If we again open our door, the molecules will spread out to achieve equal concentrations on either side of the wall. These concentration gradients will be important later, but let's return to electricity for a moment as we consider that most electrical of cells: the neuron.
All of the cells in your body, neurons included, are slightly negatively charged; this is due largely to the fact that DNA, along with many large proteins, carry a negative charge and are bound inside of every cell. This electrical imbalance is maintained by “pumps” that are constantly at work to keep sodium ions (Na+) outside of the cell and potassium ions (K+) inside of the cell; it is the natural state when a cell is at rest. This is extremely significant however, because when “excited”, a cell becomes briefly positively charged, or “depolarized”.
Here's how: like the door in the wall in our example above, “ion channels” in the cell membrane act as gates that control the flow of ions into and out of the cell. Some ion channels are sensitive to electrical charge and are called “voltage gated”: they will open when the electrical potential increases past a critical threshold, allowing specific ions to rush down the electrochemical gradient. Other channels are opened by specific signaling molecules called neurotransmitters which are released onto the cell by neighboring neurons. These two gating mechanisms, plus the notion of gradients (described above), coupled with the fact that ion channels are specific to certain ions, form the basis not just of neuronal signaling, but of every thought you've ever had and of all animate life as we know it!
As a neuron “fires,” it transmits wave of positive charge down its entire length; this spike causes all voltage-gated Na+ ion channels in the vicinity to open, and Na+ flows into the cell (down both the electrochemical and concentration gradients) causing the area inside to become positively charged, which results in even more nearby Na+ channels opening, and so on down the length of the neuron. The Na+ channels shut quickly upon opening and experience a time delay so that cannot be reactivated (thus the wave of positive charge cannot go backwards) and simultaneously the K+ channels open, letting K+ rush out of the cell and returning the patch of cell to its negative resting potential. In this way, positive charge gets passed to the next area of the membrane, and on to the next, until finally it reaches…
Licensed under CC BY-SA 3.0 via Wikimedia Commons.
The end of a neuron, called the “axon terminal,” is the part of the cell that releases chemical signals that have effects on other neurons. Once this wave of positive charge hits the end of the cell, voltage-gated calcium (Ca2+) channels open, allowing Ca2+ to rush into the cell; this influx is important because it initiates the release of neurotransmitters onto the nearby target neurons. These neurotransmitters then bind to a variety of receptors on the target neuron, opening ion channels and resulting in either excitation or inhibition of the target neuron's activity.
The main neurotransmitter responsible for stimulating other neurons is glutamate, so we will restrict our focus to its particular characteristics. It achieves its excitatory effects on the target neuron by binding to the NMDA receptor and the AMPA receptor, both of which are ion channels that allow positively charged ions through. Thus, when glutamate binds to these receptors in the target neuron, they open and allow the influx of positive charge (K+/Na+), which excites the neuron all the way down its length in precisely the way described above.
These receptors, AMPA and NMDA, will soon become very important to our discussion, but let's take a step back and consider for a moment the nervous system as a whole. We've been talking about neurons, or nerve cells, and how the flux of ions across their cell membrane, controlled by special gates and triggered by electrochemical changes, allows them to send and receive electric signals to and from other neurons. These signals typically occur at at the synapse, the area where two neurons connect, and they can be excitatory or inhibitory, increasing or decreasing the likelihood that the target neuron will itself “fire”. Multiple neurons can have the same target neuron, and many small excitatory impulses can add up to be equivalent to a single large impulse. Also, if several neurons are sending inhibitory signals while several others are sending excitatory signals, these opposing signals cancel each other out, yielding no stimulation of the target neuron. Thus, neuronal signaling depends on the summation of various positive and negative impulses over space and time; the target neuron will only “fire” if the ratio of excitatory signals to inhibitory signals heavily favors the excitatory signals.
There are around 100 billion neurons in a human nervous system forming ~$10^15$ (1 quadrillion) synapses, or interconnections. The classic neuron can take many inputs but produces only few outputs; if the inputs from other neurons sufficiently excite the neuron above a threshold value, the neuron will “fire” and a wave of positive charge will propagate down the length of the neuron to its output, causing it to release neurotransmitter onto its own target neuron, which may in turn cause that neuron to fire. These “firings” (formally called action potentials) are all the same size; you can't have a big firings and small firings; however, in the presence of increasing excitation, a neuron will fire more and more rapidly, resulting in the release of more neurotransmitter at the synapse. Similarly, two or more neurons can fire simultaneously on the same target neuron, together releasing a large amount of neurotransmitter. This high-frequency/multi-neuron stimulation is very important; more than simply exciting the target neuron and causing it to fire, this insistent stimulation strengthens the connection between these neurons. It makes the target neuron more sensitive to the specific neurons that sent the high-frequency signals. Thus, neural connections that “fire together, wire together”: these high-frequency and/or multi-neuron inputs make the target neuron more likely to fire in the future, responding to less stimulation than it took before and thus enhancing signal transmission in that specific pathway. Neuroscientist Donald Hebb is famous for proposing this sort of neuronal adaptation in his 1949 book The Organization of Behavior:
The phenomenon of long-lasting enhancement in signal transmission between neurons, known as Long Term Potentiation (LTP), is the chief mechanism underlying memory formation. It occurs all throughout the brain and can be mediated by any number of neurotransmitters, though it has been most extensively studied with glutamate release in the hippocampus. But what actually makes it work? Recall that when glutamate is released at a synapse, it binds to receptors on the target neuron and causes ion channels to open. The AMPA receptor is one such receptor; when activated by glutamate, it allows Na+ to rush into the target cell, thus increasing the positive charge inside. With small amounts of glutamate release (a small amount of stimulation), this is all that happens. The other glutamate receptor, NMDA, remains inactive.
First stage: few postsynaptic AMPA receptors (Wikimedia Commons)
But high-frequency and/or multi-neuron stimulation results in much more glutamate release, which binds to more AMPA receptors on the target neuron, which opens more gates and allows more Na+ to flow inside, which further depolarizes and excites the target cell. NMDA receptors only unlock their ion channels (1) after being activated by glutamate, and (2) when the concentration of Na+ inside the cell becomes sufficiently high. The NMDA receptors are special in that they allow both Na+ and calcium (Ca2+) to enter the cell. This is important because Ca2+ has special signaling properties once inside the target neuron; first, by activating certain proteins that insert more AMPA receptors in the membrane of the target neuron, thereby increasing it's future sensitivity to glutamate. Secondly, this influx of Ca2+ can actually initiate gene expression, resulting not only in the production of still more AMPA receptors, but also in the synthesis of proteins called growth factors which stimulate the formation of new synapses by increasing the size of the target neuron's input sites (called dendritic spines).
Simplified diagram: few postsynaptic AMPA receptors
Second stage: more AMPA receptors (Wikimedia Commons)
Third stage: more sensitivity! (Wikimedia Commons)
To recap, the increased activity between two neurons results in an increase of AMPA receptors and synaptic connections which allows future action potentials to cause a greater depolarization event in (and a greater excitation of) the target neuron. Continuous activation of the same pathways will create high-frequency action potentials and increased stimulation of the target neuron in those paths; these events strengthen the connections in a specific pathway, causing the neurons involved to become more sensitive to each other, an increased sensitivity which indicates the heightened importance of the connection.
Imagine that you live in the jungle and you have neurons that fire when they detect certain attributes of your visual perception: one may fire when black-and-orange stripes appear on your retina, another may respond to cat-like movements, etc; imagine that these neurons all begin firing at once; this special combination of firings lights up a pathway to other neurons that fire when something is threatening. Your memory that tigers are dangerous consists of the coincidence of these simultaneous rapid-fire signals on certain target neurons which embody “danger” and serve heighten your fear response, getting your body ready for the worst; the connection between these neurons will be very strong, because otherwise you would be dead already.
Now imagine that you are scooped from the jungle and deposited in a center for cat adoptions in large, tiger-free, modern industrialized city; many of your tiger-attribute detectors will be screaming bloody murder and chances are you will be freaking out. It would be extremely maladaptive if this was your reaction to all future feline encounters; there has to be a way to weaken the association between feline-detector neurons and danger neuron, some process that works in opposition to LTP. This process, called Long Term Depression (LTD), comes about through processes very similar to those that cause LTP. This weakening of synaptic strength occurs through extended periods of low-frequency stimulation; LTD occurs with small, slow influx of Ca2+ , not large enough to exceed the threshold required to recruit more AMPA receptors.
The magnitude of calcium signal in the target neuron largely governs whether LTP or LTD occurs: just as high levels of Ca2+ serve to sensitize the neuronal pathway, small levels of Ca2+ work to desensitize the pathway by removing AMPA receptors and by deactivating certain proteins. Long-term exposure to harmless kittens will lead to extended periods of low-frequency stimulation, which serves to reduce the sensitivity of danger neurons to the cat-attribute neurons (in our ad hoc, unrealistic, extremely-simplified example).
LTP and LTD are basic mechanisms of synaptic plasticity, or the ability of the brain's neuronal wiring to change over time in the presence of different circumstances conditions. In general, the idea is that if a set of inputs cause the same pattern of activity to occur repeatedly, then the active elements constituting that pattern will become increasingly strongly associated; each element in the pattern will tend to activate every other element in the pattern and to deactivate those elements that are not part of the pattern.