Memory

Mini Review

• Memory has three stages: encoding, storage, and retrieval.
• Working memory holds a few things for a few seconds. Long-term memory stores everything a person knows.
• Information reaches long-term memory more reliably through deep processing than through shallow processing.

Review: The Two Memory Systems

• So far, we have covered how information moves between the two systems. Now we are talking about what is inside long-term memory and how retrieval finds it.

Two Kinds of Long-Term Memory: Explicit Memory

• Long-term memory holds two kinds of memory, and the difference between them is whether the person knows the memory is there.
• Explicit memory: Memory that a person consciously and deliberately retrieves.
  • Ex: Remembering the name of your third-grade teacher when someone asks.
  • Ex: Recalling the definition of encoding on a quiz.
• Explicit memory is what a person means by “I remember.” Retrieving it is controlled processing: the person decides to search and knows when the search has worked.

Two Kinds of Long-Term Memory: Implicit Memory

• Implicit memory: Memory that a person is not aware of but that still influences what they feel, do, or think.
  • Ex: Typing a password without being able to say what the letters are.
  • Ex: Feeling uneasy in a hallway that smells like a hospital you stayed in as a child.
  • Ex: Humming a jingle from a commercial you do not remember watching.
• Implicit memory runs at the automatic end. It shows up in behavior with no effort, no decision to retrieve, and no awareness that anything was retrieved.

The Mirror-Drawing Study: Implicit Memory Without Explicit Memory

• The mirror-drawing study (1962): H.M., the patient from the brain deck who could not form new explicit memories after his hippocampus was removed, practiced tracing a star while seeing his hand only in a mirror, once a day for three days.
  • Finding: His tracing improved from day to day at the rate of a healthy person, and on each day he said he had never done the task before. The skill was stored as implicit memory with no explicit memory of the practice.
  • Since then: The same split has been found in every group of amnesia patients studied, and a meta-analysis (2020) of 17 studies found skill learning preserved in people with memory-impairing dementia.

Two Types of Explicit Memory: Episodic and Semantic

• Episodic memory: Memory for personally experienced events, including details of when and where they happened.
  • Ex: The afternoon you got your driver’s license.
  • Ex: What you had for dinner last night.
• Semantic memory: Memory for general knowledge that is not tied to a particular time and place.
  • Ex: Frankfort is the capital of Kentucky.
  • Ex: A driver’s license lets a person drive.

How Episodic and Semantic Memory Depend on Each Other

• Episodes add detail to concepts from the bottom up, and concepts shape how a new episode is encoded from the top down.
  • Ex: Ten visits to the DMV build a concept of what a DMV is like, and that concept decides what you notice on the eleventh visit.

The Kinds of Long-Term Memory

• Explicit memory splits by whether the memory has a when and a where. Implicit memory has no conscious retrieval to split.

Poll 1

• One word will appear on the next slide.
• Type the first word that comes to mind. Do not think about it.

Poll 1: First Word That Comes to Mind

Poll 2

• Same task, one more word.

Poll 2: First Word That Comes to Mind

Poll 1 Results

Poll 2 Results

Semantic Memory as a Network of Connected Concepts

• Semantic memory is organized as a network. Each concept is a node, and each association between two concepts is a link.
  • Ex: The concept of a nurse is linked to hospital, doctor, patient, and uniform.
• A link is a neural pathway one level up. Two concepts that are used together often are carried by pathways that fire together, and repeated use makes those pathways carry the signal more easily.
• When one node becomes active, activation spreads along its links, and the linked nodes become easier to reach. The polls you just answered show which links are strongest for the whole class.

The Word-Pair Study: Activation Spreading Through the Network

• The results show which words the class produced most often for each cue. Those words are the nodes with the strongest links to the cue.
• The word-pair study (1971): People saw two strings of letters and pressed a key as fast as they could to say whether both were real words.
  • Finding: People answered faster for related pairs such as NURSE and DOCTOR than for unrelated pairs such as NURSE and BUTTER. Reading the first word activated its node, and activation had already spread to the second word before it appeared.
  • Since then: A four-university study (2013) collected the same task from 768 people on 1,661 words and found the effect throughout, and it is one of the most repeated findings in cognitive psychology.

Why Retrieval Cues Work

• Retrieval cue: Anything present at retrieval that helps bring a memory back.
• A retrieval cue works the same way as the words in the polls. The cue activates a node near the one you are looking for, activation spreads along the link, and the target becomes reachable.
  • Ex: You cannot remember an actor’s name. Someone says “she was in that vampire movie,” and the name arrives.
• A concept with many links can be reached from many directions. A concept with one link can be reached from one direction.

Our Class Site as a Memory Network

• The class site is built the way semantic memory is built.
  • A key-term page is a node. Each term has one page and one definition.
  • A link between two pages is an association. Hovering over a link shows the linked definition, which is activation spreading to a neighbor.
  • The graph on each page is that node’s neighborhood, and the graph of the whole site is the network of the course.

Watching a Memory Network Grow

Live demonstration

kendalsmith.com/cognitive/key-terms/encoding

Many Cues to One Concept

• On the site, “thoughts, feelings, or behaviors,” “what a person feels, does, and thinks,” and “the ABCs” all open the same page.
• In memory, the same holds. A concept you have met under several wordings has several cues that lead to it, and a concept you have met under one wording has one.
• Each time a term appears in a slide, a flashcard, or another course, a new link is added to the same node.
• Half of you are in PSY 317 as well, and correlation and attention are nodes that both courses link.

What the Network Means for Studying

• A definition read once is a node with one link.
• Writing the definition in your own words, adding an example, and connecting it to another term add links to that node, which is deep processing from Memory Systems.
• Part 4 of this unit returns to the network to explain why some study methods work and others do not.

Retrieval Cues

• Retrieval is a search for a node, and the search starts from whatever is present at the moment.
• A retrieval cue can be a word, a place, a smell, a song, or a feeling.
  • Ex: A song from the summer you learned to drive brings back that summer.
  • Ex: The smell of a hospital brings back a night spent in one.
• Retrieval works best when the cues present now match the cues that were present when the memory was made.

The Diver Study: Matching the Cues at Encoding

• The diver study (1975): Divers learned a word list either on land or underwater, and then recalled it either on land or underwater.
  • Finding: Words learned underwater were recalled better underwater, and words learned on land were recalled better on land. The setting itself was a cue.
  • Since then: A meta-analysis (2001) of the studies on physical setting found a real but small effect that appears for recall and disappears for recognition. A direct replication with 16 divers (2021), run in one day in a heated indoor pool, did not find the effect. The setting is one cue among many, and it matters most when no stronger cue is available.

Re-Creating Retrieval Cues on Purpose

• The cues do not have to be physically present. Imagining them activates the same nodes.
• Re-creating the setting of encoding in your head is a study strategy, and witnesses are questioned the same way.
  • Ex: Before answering a question about the brain deck, you picture the room, the slide with the four lobes, and where you were sitting, and the answer comes.
  • Ex: A witness is asked to think back to the weather, the sounds, and what they were feeling before being asked what they saw.

Poll 3

• Twelve words will appear one at a time. Read each one and try to remember it.
• Do not write anything down.
• When the list ends, type every word you remember into the poll.

Poll 3: Which Words Do You Remember?

Poll 3 Results

Poll 4

• Eighteen words are listed in the next poll. Twelve were on the list you just saw and six were not.
• For each word, rate how sure you are that it was on the list.

Poll 4: Which Words Were on the List?

Poll 4 Results

Two Ways to Test Memory: Recall and Recognition

• Recall: Retrieving information from memory with few or no retrieval cues.
  • Ex: Writing the definition of a key term on a quiz.
  • Ex: Poll 3, where the only cue was “the list you just saw.”
  • Ex: Describing to a police officer what the driver who hit your car looked like.
• Recognition: Retrieving information from memory when the information itself is present as the retrieval cue.
  • Ex: Answering a multiple-choice or matching question.
  • Ex: Poll 4, where every word was on the screen.
  • Ex: Pointing out that same driver in a police lineup.
• Compare the counts on the Poll 3 results with the ratings on the Poll 4 results, word by word.
• Recognition is easier because every item presented is a cue, and the cue activates the node directly.

Why Quizzes Ask You to Write the Definition

• Recognition can be passed on familiarity alone. You might be able to select the correct definition from a list, but that does not show whether you could have produced it. Why does that matter?
• Producing a definition from the term alone requires deep processing, and each successful retrieval strengthens the path that retrieval will follow next time, as well as expanding the network that can activate its retrieval.
• Picking the definition from a list relies on shallow processing and makes it less likely that you will be able to retrieve the concept in the real world.

Source Memory

• Source memory is memory for where and when something was learned.
• Nearly everyone knows a spider has eight legs, and almost no one can say where the fact came from. The fact is in semantic memory, but the episode that put it there is gone.
• Recognition has two parts. Familiarity is the fast sense that something has been encountered before. Source memory is knowing where.
• Familiarity without source memory is where errors start.
  • Ex: A name feels familiar, and a person decides it belongs to someone famous when it was on a list they saw last week.
• We will talk more about those errors in Memory Errors and Forgetting.