Long-term memory exists to store information for later retrieval. However, when we retrieve information from memory, we are not getting a replica, but rather a reconstruction. Our memories are fluid, not static, meaning that they change as a result of our current beliefs and perceptions about the past, present, and future. A memory is not simply a photograph or audio clip, but rather becomes colored by our thoughts and feelings at the time, and our current beliefs about what the memory should represent.

Long-term memories reside in a complex network of associations to other memories, thoughts, and feelings, similar to a spider's web. Unlike short-term memory, long-term memory seems to have an unlimited capacity, never becoming "full" of information (at least, we haven't found its limit yet). There is some loss of information over time, but it is difficult to determine whether this loss is due to decay or interference, or whether the information is in fact still there, but we are unable to retrieve it. The most important thing to understand about long-term memory is that individual memories are not filed away into separate boxes, but rather are connected to other memories by a network of associations. Activating one memory can, by association, can call up another related memory. We can see this network in action in the way that people encode and retrieve items from long-term memory. For example, given a list of words, some of which are related to each other (e.g. salt/pepper, hot/cold, mother/father), people tend to recall the related words together, no matter how far apart they were spaced on the original list.

Encoding is the process of entering information into long-term memory. We measure the efficiency of encoding a particular item by how easy it is to retrieve that item from memory later. For example, if I study a painting intently, encoding it into my memory, and later I find that I can easily recall many details of the artwork, then I can say that I efficiently encoded that painting. In general, having more possible paths of associations to an item in memory will make it easier to reach that item later, upon retrieval.

One way that people try to encode information is through rehearsal. Simply studying an item repeatedly can improve memory. For example, if I want to memorize the first line of the Gettysburg Address, I might repeat "Four score and seven years ago..." over and over again until it has become almost effortless. Spacing out such rehearsal over time can also help encoding efficiency. If I only repeat the line fifty times today, I may not remember it in a week. But if I repeat it ten times a day for five days, I will be able to recite the line much more easily later. This is called the spacing effect.

In addition to simply increasing the number of times we encode an item, we can also change the way we encode it. There are two principal types of encoding: deep and shallow. Shallow encoding focuses on the superficial aspects of the item, such as the number of vowels or the sound of the words. Rehearsal, simply repeating the words over and over, is a form of shallow encoding. Deep encoding focuses on drawing connections between that item and other information already in memory. Deep encoding takes more effort, but it also improves encoding efficiency. If I wanted to encode a new vocabulary word deeply, for example, I might try to make a sentence using the word, or think about what similar words I already know. By encoding an item deeply in long- term memory, we create more connections between it and other items in memory. The increased number of associations make it more likely that we can find it by tracing a path through memory.

Retrieval is the process of digging information out of memory. When I ask you, "What is the capital of Thailand?" or "Do you know the woman in this photograph?", I am asking you to retrieve that information from storage in your long-term memory. There are two principal types of retrieval: recall and recognition. "What is the capital of Thailand?" is a question that requires recall. You must somehow use the cues "capital" and "Thailand" to find the answer, "Bangkok." On the other hand, the question "Do you know the woman in this photograph?" requires recognition. You only need to decide whether she is familiar, whether she exists already as an item in your long-term memory. In general, recognition is easier than recall.

In the web-like network of long-term memory, retrieval is a search along the strands of the web. You can take many different paths to find the particular item that you need, because, since each item is associated with many others and so is touched by many strands. The more strands that touch the item you seek, the easier it is to find that item, since more possible paths will lead to it. For example, if you think of the word "sleep," you probably have many associations with it: going to bed, feeling tired, the nap you took yesterday, your brother's snore, etc. If I tell you to think of bed, or if you feel tired, you'll probably think of "sleep." This is basically how retrieval works. It uses existing associations to trace from a given starting point, called a cue, back to other memories, and bring them into awareness.

For specific, event-related memories, the context in which the memory was learned is also included in associations, and it can be used as a cue to retrieve the memory later. This is called context-dependent memory. In one well-known experiment, subjects studied a list of words while scuba diving underwater. Later, they were able to recall more of the words if they were underwater again than if they were on dry land. Recreating the context activates one more association to the sought-after items, making it more likely that they will be recalled.

Mnemonic devices attempt to improve encoding specificity by using tricks to deeply encode items in long-term memory with very little effort. One well- know mnemonic device is first-letter conversion. In first-letter conversion, you take the first letter of each item on a list and use them to spell out a short word or phrase. For example, the name ROY G. BIV stands for the colors of the rainbow: Red, Orange, Yellow, Green, Blue, Indigo, and Violet. This technique is easy to use and more efficient than straight memorization, but not as effective as other mnemonics. The peg word technique uses visual imagery paired with key rhyming phrases. A commonly used rhyming phrase is, "one is a bun, two is a shoe, three is a tree…" and so on. If I want to remember what groceries I need to buy at the store, I create a series of mental images that pair each grocery item with the numbered item in the phrase. For example, if I need to buy soap, juice, and cookies, I could mentally picture a sandwich made of soap on a bun ("one is a bun"), a shoe filled with juice ("two is a shoe"), and a cookies hanging from a tree ("three is a tree"). Although the peg word technique can be helpful for a while, if you use it too often, new items tend to get confused with images from past lists. For example, if I need to buy milk at the store next week, I could picture a shoe filled with milk, but, once I get to the store, I might mistakenly remember the juice-filled shoe instead.

A third mnemonic device that gets around this problem, yet is highly effective, is the method of loci ("loci" is a Greek word meaning "places," and this mnemonic device was originally used by Greek orators). Imagine a route you know well, such as the trip from your room to the kitchen in your house. Choose several spots along that route, for example, the doorknob of your room, the top of the stairs, the table in the hallway, etc. Now, imagine yourself walking along that route and placing the items you want to remember in those chosen places. For example, I could balance a bar of soap on my doorknob, place a carton of juice at the top of the stairs, and leave a plate of cookies on the table downstairs. All you need to do when you want to remember the items is mentally walk that route, and you will see each item in your mind's eye as you reach each chosen spot. For each new list, just imagine a new route, such as the path from your house to your school, and you will avoid the confusion that can interfere with retrieving memories in the peg word method.