Memory feels like a recording: something happened, we stored it, and we play it back. Psychology has shown that this picture is wrong in important ways. Memory is less like a video library and more like a reconstruction, rebuilt each time we recall it from fragments, expectations and later information. Understanding how it works helps us study better, judge eyewitness testimony more wisely and be more humble about our own certainty.

The forgetting curve

The first scientific memory experiments were done by the German psychologist Hermann Ebbinghaus, who published his results in 1885. To avoid the influence of meaning, he memorized long lists of nonsense syllables such as "DAX" and "BOK," then tested himself after different delays. He found that forgetting is fastest at first and then levels off, a pattern now called the forgetting curve. He also found that relearning a list was quicker than learning it the first time, which means that "forgotten" memories often leave traces behind.

Stages of memory

A classic 1968 model by Richard Atkinson and Richard Shiffrin describes three stages:

• Sensory memory holds a brief snapshot of what the senses take in, lasting only a fraction of a second for vision.
• Short-term memory holds what we are currently thinking about, for perhaps twenty seconds unless we rehearse it.
• Long-term memory stores information for hours, years or a lifetime, with no known limit on capacity.

Short-term memory is small. In 1956 George Miller wrote a famous paper on "the magical number seven, plus or minus two," and later research suggests the true limit may be closer to four meaningful chunks. Chunking is why phone numbers are grouped: 5-5-5-0-1-9-9 is harder to hold than 555-0199. In 1974 Alan Baddeley and Graham Hitch refined the idea into working memory, an active mental workspace where we hold and manipulate information, such as when doing arithmetic in our heads.

Different kinds of long-term memory

Long-term memory is not one system. Explicit memories are those we can consciously recall, including facts (Paris is the capital of France) and personal episodes (my tenth birthday). Implicit memories show up in what we do rather than what we can say, such as how to ride a bicycle. The clearest evidence for this split came from a patient known for decades as H.M., Henry Molaison. In 1953, surgeons removed parts of both of his temporal lobes, including most of the hippocampus, to treat severe epilepsy. Afterward he could not form new explicit memories: he would meet the same researcher again and again as a stranger. Yet when the psychologist Brenda Milner taught him to trace a star while looking in a mirror, he improved over several days, while insisting he had never done the task before.

Memory is reconstructive

The American psychologist Elizabeth Loftus has spent decades showing how easily memories can be changed. In a 1974 study with John Palmer, volunteers watched films of car crashes and were asked how fast the cars were going when they "hit" or "smashed into" each other. Those asked with the word "smashed" gave higher speed estimates. A week later, they were also more than twice as likely to remember seeing broken glass, though there was none in the film. Later studies went further: by suggesting false childhood events, such as being lost in a shopping mall, researchers led about a quarter of participants to recall at least part of an event that never happened.

This has serious consequences. Mistaken eyewitness identification has been a factor in a large majority of the wrongful convictions later overturned by DNA evidence in the United States. Confidence is not a reliable guide either, since a memory can feel vivid and certain and still be wrong, especially after it has been retold many times. Our tendency to remember things in ways that fit what we already believe is one of many mental shortcuts known as cognitive biases.

What helps memory

Research points to a few strategies that work far better than rereading notes:

• Retrieval practice: testing yourself, which strengthens memories more than reviewing them.
• Spacing: spreading study over several sessions instead of cramming.
• Elaboration: connecting new information to what you already know and asking why it is true.
• Sleep: during sleep the brain consolidates recent learning into longer-lasting form.

Worked example: studying for an exam

Two students have a week before a biology exam. One reads the textbook chapters three times the night before, highlighting heavily. The other spends half an hour each day closing the book and writing down everything they can remember, checking what they missed, and explaining the ideas to a roommate. The first student will probably feel more confident, because rereading makes material feel familiar. The second student will probably remember far more, because each act of retrieval strengthens the memory and each day's gap fights the forgetting curve. Ebbinghaus would not be surprised. Familiarity is not the same as knowing, and the most useful study habits are the ones that feel a little effortful.