Lecture: Memory
Lecture script · about 15–20 minutes spoken · Week 7
Opening
A nursing student learns a long list of drug names for a licensing exam. A juror listens to an eyewitness who is completely sure. A new warehouse hire has to hold a ten-digit order code in mind long enough to type it. All three are leaning on the same system, and all three can be let down by it. This lecture is about that system: how memories are made, why they fade, and why a memory you're sure of can still be wrong.
You've been building toward this. In Week 2, you met two studies as models of method: Ebbinghaus measuring memory as percent savings, and Loftus and Palmer changing one verb in a question. In Week 3, you met the hippocampus and the case of H.M. In Week 6, you saw how experience changes behavior. Learning is the change. Memory is how the change is kept. Next week, in Week 8, you'll review all of it for the midterm.
Here's the plan: three processes, three stages, a small workspace, what makes memories stick, kinds of memory, forgetting, rebuilding, and the brain. If you keep one sentence, keep this one: memory is rebuilt, not replayed.
1. Follow a memory through three processes: encoding, storage, retrieval
Picture a server at a busy restaurant taking an order with no notepad. First the server has to take in "no onions, dressing on the side." Then the order has to stay put while the server walks past three other tables. Then it has to come back out at the kitchen window. Three steps, and a mistake can happen at any one of them.
Psychologists name those steps, always in this order. Encoding is the first process of memory: turning what you sense or think into a form the brain can store, such as noticing a name and linking it to a face. Storage is the second process of memory: keeping encoded information over time, from a fraction of a second to a lifetime. Retrieval is the third process of memory: getting stored information back into awareness when you need it, such as recalling an answer during an exam.
That changes what "I forgot" means. Forgetting is the inability to retrieve information that was once available, which can come from a failure of encoding, storage, or retrieval. Think of a classmate's name that sits on the tip of your tongue all afternoon and then shows up while you're brushing your teeth. Was it lost? No. It was there the whole time. You couldn't reach it yet.
Pause and try this. You're introduced to someone while reading a text, and an hour later you have no idea what their name was. Which of the three processes failed?
Think about where your attention was.
Here's the answer. Encoding failed, because you never paid attention to the name, so it never got into memory to begin with.
2. Trace the three-stage model: sensory, short-term, and long-term memory
You're reading on the couch and someone across the room says something. You say "What?" and then, before they repeat it, you hear the sentence in your head. For a second or two, the sound waited for you. That echo is the first stage.
The three-stage model is the model of memory, proposed by Atkinson and Shiffrin, in which information passes from sensory memory to short-term memory and, if it is rehearsed or encoded, into long-term memory. Sensory memory holds a brief, nearly complete copy of what your senses take in for about a second or two. Short-term memory holds a small amount of information, about seven items give or take two, for about 15 to 30 seconds unless it is rehearsed. Long-term memory is a relatively permanent store with no known limit on how much it can hold.
Those short-term numbers come from classic work. Miller, in 1956, called the span "the magical number seven": 7 plus or minus 2 items, so somewhere from 5 to 9. Peterson and Peterson, in 1959, showed how short the hold is. They gave people a trio of letters, then had them count backward so they couldn't rehearse. Recall was about 80% after 3 seconds and about 10% after 18 seconds.
Pause and try this. A friend says short-term memory is what lets you remember what you ate for breakfast. Using the pinned duration, what's wrong with that?
Hold the number 30 in mind.
Here's the answer. Breakfast was hours ago, and short-term memory holds information for only about 15 to 30 seconds without rehearsal, so breakfast lives in long-term memory.
3. Use working memory: chunking and the limit of about four
Split a dinner bill four ways in your head and notice what you're doing. You hold the total, divide it, keep the answer, and add the tip. You're not storing a number. You're working on it.
That's why psychologists now talk about working memory: short-term memory understood as an active workspace, where you hold and manipulate a few pieces of information at once. Cowan, in 2010, put its capacity at about 4 plus or minus 1 chunks, so from 3 to 5. That's smaller than Miller's 7, because Cowan counts chunks, not raw items.
Here's the trick that makes a small space go further. Chunking is grouping separate items into larger, meaningful units, such as reading ten digits as three groups, so that more information fits into the limited space of short-term memory. Take a ten-digit phone number. As single digits, it's 10 items, more than the upper limit of 5. As three groups of 3, 3, and 4 digits, it's 3 chunks, comfortably inside the range. Chunking doesn't make the workspace bigger. It makes each piece bigger.
Pause and try this. Miller's span is 7 plus or minus 2 items, and Cowan's capacity is 4 plus or minus 1 chunks. Which estimate is smaller, and why don't the two contradict each other?
Think about what each one is counting.
Here's the answer. Cowan's is smaller, and they don't clash, because Miller counted raw items while Cowan counts chunks held in an active workspace.
4. Encode for keeps: deep processing, elaboration, and spacing
Two roommates study the same vocabulary list. One reads it aloud ten times. The other asks what each word means, ties each one to something familiar, and quizzes herself on three different days. A week later, the second roommate remembers far more. Both worked. Only one worked on meaning.
Rote repetition is repeating information over and over without thinking about its meaning; it can hold an item in short-term memory but encodes it weakly for the long term. Deep processing is encoding that focuses on the meaning of information rather than on its surface features. Elaborative rehearsal is encoding new information by linking it to its meaning and to things you already know. And the spacing effect is the finding that study spread over several shorter sessions produces better long-term retention than the same amount of study packed into one long session.
Here's an analogy. Cramming is like watering a plant with a whole bucket once a week. Most of it runs off. Spacing is a cup a day, and the soil actually takes it in.
Pause and try this. Two students each study for the same total time. One does it all the night before; the other splits it across three days. Which one does the spacing effect favor?
Notice that the total time is the same.
Here's the answer. The student who splits it across three days, because spreading the same amount of study over shorter sessions produces better long-term retention.
5. Sort your memories: explicit memory versus implicit memory
Think about a weekend soccer player. She can tell you about the rainy away game when she scored from a corner kick. She knows the offside rule. And she can dribble past a defender without thinking about her feet. Those are three different kinds of long-term memory.
Explicit memory is memory a person can consciously recall and put into words, also called declarative memory. It comes in two kinds. Episodic memory is explicit memory for personal experiences tied to a particular time and place, like that rainy game. Semantic memory is explicit memory for general facts and knowledge, like the offside rule, without remembering when you learned it. Implicit memory is memory that shows up in what a person does rather than in what they can consciously recall, also called nondeclarative memory. Its main kind is procedural memory: implicit memory for skills and habits, such as riding a bike or typing, that you perform without consciously recalling the steps.
A quick test helps. Can you say it, and is it about your own moment? Then it's episodic. Can you say it, but it's general knowledge? Semantic. Can you do it without being able to say how? Procedural.
Pause and try this. You type your phone's passcode without looking, but when a friend asks for it, you have to watch your own thumbs to tell them. What kind of memory is doing the typing?
Ask whether you could say it or only do it.
Here's the answer. Procedural memory, a kind of implicit memory, because the knowledge shows up in what your thumbs do rather than in what you can recall and say.
6. Measure forgetting: the forgetting curve and interference
In 1885, Ebbinghaus wanted to measure even the memory he could no longer recall. His answer was percent savings: how much less time relearning a list took than learning it the first time. Savings catch memory that's hidden but still there.
Here's what he found. Picture a line that starts high on the left and drops steeply, then bends and runs almost flat to the right. Savings were 58.2% after 19 minutes and 21.1% after 31 days, a total drop of 37.1 points. But look at where the drop happens. From 19 minutes to 63 minutes, savings fell from 58.2% to 44.2%, a drop of 14.0 points in 44 minutes. From 6 days to 31 days, they fell from 25.4% to 21.1%, only 4.3 points across 25 days. That shape is the forgetting curve: retention drops steeply soon after learning and then levels off, so most forgetting happens early.
Why does memory fade? Sometimes it was never encoded. Sometimes it's stored but out of reach. And sometimes other memories get in the way. Proactive interference is older learning getting in the way of recalling newer learning, like writing your old phone number when you mean to give your new one. Retroactive interference is newer learning getting in the way of recalling older learning, like being unable to recall last year's locker combination after learning this year's. Here's a hook: pro means forward, so the old pushes forward into the new; retro means backward, so the new reaches back into the old.
Pause and try this. You've learned a new route to work, and now, when a friend asks how you used to go, you can only think of the new route. Proactive or retroactive?
Ask which memory you're trying to reach.
Here's the answer. Retroactive interference, because you're trying to retrieve the older route and the newer learning is blocking it.
7. Rebuild, don't replay: reconstructive memory and the misinformation effect
Ask two friends to describe the same concert a year later. Both will be sure. Their stories won't match. Neither is lying. Both are rebuilding. Reconstructive memory is the idea that recall rebuilds a memory from stored pieces, expectations, and later information each time, rather than replaying an exact recording.
Loftus and Palmer, in 1974, showed how far the rebuilding goes. Students watched films of car accidents, then were asked how fast the cars were going when they smashed, collided, bumped, hit, or contacted each other. The verb was the independent variable, and the speed estimate was the dependent variable. Smashed averaged 40.8 miles per hour; contacted averaged 31.8, a gap of 9.0. A week later, 16 of 50 students in the smashed group, or 32%, reported broken glass. There was no broken glass in the film. Compare that with 14% for hit and 12% for students asked no speed question at all. That's about one student in three remembering glass that wasn't there. Because students were randomly assigned to a verb, the design lets you say the wording changed the memory.
That's the misinformation effect: misleading information supplied after an event can change what a person later remembers about it. It's true even for a flashbulb memory, a vivid, detailed memory of the moment you learned surprising, emotional news. Those feel exact. They can still change.
Pause and try this. An eyewitness is very confident about a detail. Based on Loftus and Palmer, should that confidence settle the question?
Think about the students and the broken glass.
Here's the answer. No, because confidence tells you how a memory feels, not whether it matches what happened; students confidently reported glass that was never in the film.
8. Locate memory in the brain: hippocampus, amygdala, cerebellum, and H.M.
Picture a cyclist who nearly gets hit by a car turning without signaling. Years later she remembers the moment in sharp detail, and she brakes smoothly on that corner without thinking. Different parts of her brain hold different parts of that day.
You met these structures in Week 3. The hippocampus is needed to form new explicit memories, the facts and events you can consciously recall. The amygdala is central to fear and strengthens memories carrying emotional weight, which is part of why the near-miss stays vivid. The cerebellum coordinates movement and supports learning motor skills, like that smooth braking.
The case of H.M., reported by Scoville and Milner in 1957, shows in one person that new explicit memories and new motor skills rely on different systems. To control seizures, surgeons removed much of the medial temporal lobe on both sides, including much of both hippocampi. Afterward, he could no longer form new declarative memories. He kept most memories from before the operation. And he could still learn new motor skills. One caution: H.M. is a case study, one person, so his case suggests these links rather than proving them for everyone.
Pause and try this. H.M. could still learn new motor skills. Which kind of memory did that show was still working?
Think back to the soccer player's feet.
Here's the answer. Procedural memory, the implicit memory for skills, which doesn't depend on the hippocampus in the way new explicit memories do.
Closing
Let me pull it together in the order we built it. Memory runs in three processes: encoding, storage, and retrieval, and forgetting can happen at any of them. Sensory memory lasts a second or two; short-term memory holds about 7 plus or minus 2 items for about 15 to 30 seconds. Working memory is an active workspace of about 4 chunks, and chunking fits more into it. Deep processing, elaboration, and spacing beat rote repetition and cramming. Explicit memory is episodic or semantic; implicit memory includes procedural skills. The forgetting curve drops steeply, then levels off, and interference runs both ways. Memory is rebuilt, and confidence is not accuracy. And the hippocampus, amygdala, and cerebellum each support a different kind of memory.
Next in this module come the slide deck, the Readings & Resources page, the lecture tutorial, the practice set, and the quiz, then the discussion, the assignment, and the Analysis Workshop. Next week, in Week 8, Midterm Review & Exam, you'll use the "how do we know?" checklist on Loftus and Palmer and on the sleep and quiz-score data you met earlier.
Built with The Course Maker · samples.thecoursemaker.com