Memory

• 1. Introduction

  ›

  Memory is the faculty of the mind by which data or information is encoded, stored, and retrieved when needed. Without it, the source argues, language, relationships, and personal identity could not develop. Forgetting is the everyday name for its failure. A disorder such as amnesia is the clinical one. The source frames memory as an information processing system, built from a sensory processor, short-term or working memory, and long-term memory. Yet that tidy architecture hides a stranger truth. Memory is not recording equipment. It is constructed when we encode an event, and reconstructed again every time we recall it. How long can a single flicker of sight survive? Why does the wording of a question reshape what a witness swears they saw? And how can a plant with no brain at all remember a touch it felt thirty seconds ago? The chapters that follow trace memory from milliseconds to a lifetime, and from the human hippocampus to the leaf of a Venus flytrap.

• 2. Stores And Sketchpads

  ›

  George Sperling, in 1963, flashed a grid of twelve letters at subjects, three rows of four, then played a high, medium, or low tone telling them which row to report. Using this partial report paradigm, he showed that sensory memory could hold roughly twelve items but decayed within a few hundred milliseconds. This store splits into three kinds. Iconic memory holds visual information, echoic memory holds auditory information, and haptic memory holds touch. In 1956, George A. Miller, working at Bell Laboratories, put the limit of short-term memory at seven plus or minus two items. His famous paper carried that figure in its title. Modern perspectives revise the number downward, to roughly four or five items, or argue for a flexible limit based on information rather than items. Chunking stretches the limit. A ten-digit telephone number splits into an area code, a three-digit group, and a four-digit group, which is why some countries print numbers in clusters. In 1974 Baddeley and Hitch replaced the flat idea of short-term memory with a working memory model of three stores: a central executive, a phonological loop, and a visuospatial sketchpad. The phonological loop silently rehearses sounds in a continuous loop, and is disrupted by irrelevant speech or by the word-length effect. The visuospatial sketchpad handles spatial tasks like judging distances, and people with aphantasia cannot engage it. In 2000 the model gained an episodic buffer that links visual, spatial, and verbal information into integrated units, the kind of binding that lets you hold a story or a movie scene together.

• 3. What The Brain Keeps

  ›

  Baddeley, in 1966, found that after twenty minutes subjects struggled most to recall words with similar meanings, such as big, large, great, and huge. This pointed to a key contrast. Short-term memory leans on an acoustic code, a fact Conrad had shown in 1964 with letters like E, P, and D that sounded alike. Long-term memory instead encodes meaning, encoding information semantically. Long-term memory has subcategories of its own. Anderson, in 1976, divided it into declarative or explicit memory, which requires conscious recall, and procedural or implicit memory, which does not. Declarative memory holds semantic memory, the storehouse of facts like "Paris is the capital of France," and episodic memory, which captures the what, when, and where of personal events. Episodic memories often mark the firsts in life, a first kiss, a first day of school. Procedural memory is remembering how to do something, like riding a bike or tying shoelaces, and it leans on the cerebellum and basal ganglia. Topographical memory lets a person orient in space and follow a route, so getting lost while traveling alone is its failure. Flashbulb memories are vivid episodic memories of unique, highly emotional events, such as where a person was on hearing of President Kennedy's assassination, the September 11 attacks, or the Sydney Siege. John Meacham added a different axis in a 1975 paper, later included by Ulric Neisser in his 1982 volume Memory Observed. He split memory by temporal direction into retrospective memory of the past and prospective memory of the future, the latter a matter of remembering to remember, with cues like sticky notes or a string around the finger.

• 4. Lessons Of Damaged Brains

  ›

  Patient Henry Molaison was long thought to have had both hippocampi fully removed, and from his case came the theory that the hippocampus is essential for moving new memories from short-term to long-term storage. The hippocampus is essential to consolidation, though it does not seem to store information itself, and may keep changing neural connections for three months or more after learning. A later twist complicated the story. A post-mortem examination of Molaison's brain showed the hippocampus was more intact than first believed, throwing the original conclusions into question. Patient KF, studied by Zlonoga and Gerber in 1986, broke the Atkinson-Shiffrin model in a different way. Brain-damaged, KF struggled to recognize spoken numbers, letters, words, and noises like doorbells or a cat meowing, yet visual short-term memory was unaffected, dividing visual from auditory memory. Korsakoff's syndrome, an organic brain disease, harms memory through widespread loss of neurons in the prefrontal cortex. Hyperthymesia runs the other way, leaving a person unable to forget small details that would normally vanish. Hippocampal damage can cause retrograde amnesia, the loss of memory for events just before the injury. Memory dysfunction can also follow viral infection. Many patients recovering from COVID-19 report memory lapses, and SARS-CoV-1, MERS-CoV, Ebola virus, and even influenza virus have been linked to the same trouble.

• 5. The Molecular Trace

  ›

  Long-term memory, unlike short-term memory, depends on the synthesis of new proteins. In hippocampal cells the trigger is calcium release, which follows the expulsion of magnesium that normally blocks NMDA receptors, freeing them to admit calcium and set off gene transcription. One newly made protein, an autonomously active form of protein kinase C called PKMzeta, sustains the enhancement of synaptic strength. Inhibiting PKMzeta erases established long-term memories without touching short-term memory, and once the inhibitor is gone, the ability to form new long-term memories returns. By 2015 it had become clear that long-term memory requires gene transcription activation and de novo protein synthesis. Formation depends on both activating memory-promoting genes and inhibiting memory-suppressor genes, with DNA methylation and demethylation as a major mechanism for this dual control. Rats given a strong long-term memory through contextual fear conditioning showed altered expression of 9.17 percent of the hippocampal genome a day after training, with about 1,000 genes reduced and about 500 increased. A learning experience can rapidly trigger DNA topoisomerase 2-beta, or TOP2B, to cut double-strand breaks in the promoters of immediate early genes. Four enzymes of the non-homologous end joining repair pathway, DNA-PKcs, KU70, KU80, and DNA LIGASE IV, mend those breaks within about fifteen minutes to two hours. The repair is associated with demethylation that allows the genes to switch on. CREB, a transcription factor thought important for consolidating short-term into long-term memory, is believed to be downregulated in Alzheimer's disease.

• 6. Consolidation And Sleep

  ›

  The transition of a memory from short term to long term is called memory consolidation, and the source admits little is known about the physiology. Two propositions compete. One is backpropagation, used by Geoffrey E. Hinton, a Nobel Prize laureate for Physics in 2024, to build software, though empirical evidence for it in the brain is not available. The other is positive feedback from the endocrine system. The endocrine route has experimental support. It was shown first in insects, which use arginine, nitric oxide, and endorphin receptors, and the role of arginine and nitric oxide has since been confirmed in birds, mammals, and humans. Glial cells also matter, though how remains unknown. Sleep enters the same story. System consolidation takes place during slow-wave sleep, when the hippocampus replays the day's events for the neocortex, which then moves them into long-term memory. Too little sleep leaves these neural connections weaker and learning less efficient, and sleep deprivation has even been shown to produce false memories. Neuroimaging has caught the sleeping brain in activation patterns that mirror the previous day's learning, suggesting new memories are solidified through this rehearsal. The traditional consolidation dogma has lately been re-evaluated through reconsolidation, the finding that a memory retrieved into working memory becomes open to change again, so that retrieved memory is not a carbon copy of the original experience.

• 7. Memory As Construction

  ›

  Elizabeth Loftus and John Palmer, in 1974, showed people a film of a traffic accident, then asked how fast the cars were going. Those asked how fast the cars were going when they smashed into each other gave higher speeds than those asked when the cars hit each other. A week later, the smashed group was twice as likely to report seeing broken glass, though the film showed none. The wording had reshaped the memory itself, a result replicated worldwide and named the misinformation effect. Imagination can plant memories from nothing. Goff and Roediger, in 1998, found that people who repeatedly imagined an act, like breaking a toothpick, grew more likely to believe they had actually performed it. Garry and colleagues, in 1996, asked college students to imagine four childhood events, and a fourth of them came to report the imagined events as real. False memories can even be installed by hand. Using optogenetics, a team of RIKEN-MIT scientists led mice to wrongly associate a harmless environment with an earlier unpleasant experience from different surroundings. Some scientists believe the work may bear on false memory in humans and on treating PTSD and schizophrenia. The constructive view reframes the point of remembering. Current hypotheses suggest these processes let people simulate and imagine future episodes, drawing on the past flexibly because the future never exactly repeats it.

• 8. Stress, Age, And Plants

  ›

  L. Schwabe and O. Wolf, German cognitive psychologists, immersed students' hands in ice-cold water for up to three minutes in the Socially Evaluated Cold Pressor Test, then had them memorize thirty-two words. A day later the stressed group recalled thirty percent fewer words than controls. A second Schwabe and Wolf study softened the blow. Using a card game scented with vanilla, they found that stressed subjects showed no impairment when tested in the same vanilla-scented room, because odour is a strong cue for memory, with implications for education, eyewitness testimony, and psychotherapy. Stress also leaves lasting marks. Glucocorticoids released during stress damage CA1 neurons in the hippocampus, so repeated stressful situations raise the risk of later memory loss, a chain that may help explain weaker memory performance among children from lower socioeconomic backgrounds. Age tells a parallel story. Gene transcription profiles of the human frontal cortex from ages 26 to 106 showed numerous genes with reduced expression after age 40, and especially after 70, alongside a marked rise in likely oxidative DNA damage in those genes' promoters. Infancy sits at the other end. Research now shows infants as young as six months can recall information after a 24-hour delay, with twenty-month-olds recalling action sequences twelve months later, while infantile amnesia names the accelerated forgetting of early life. The strangest case has no brain at all. The Venus flytrap, native to the subtropical wetlands of the eastern United States, snaps shut only when two of its trigger hairs are stimulated within thirty seconds. It remembers the first touch not in neurons but in cytoplasmic calcium, where a subthreshold influx waits for a second influx to superimpose, cross threshold, and close the trap.