PSYC 105 - Cog. Psych. - Midterm 2, Lowe @ UCSD

# of Stages of Memory

3

Stages of Memory

Encoding, Storage, Retrieval

Encoding

...Incoming information is acquired and processed

Storage

...A record of the encoded information is stored in the brain

Retrieval

...Information is recovered from memory stores and brought to consciousness, or in procedural memory to perform a behavior or motor act.

Developed the modal model of memory

Atkinson & Shiffrin (1968)

Modal model of memory

- Original model based on our understanding of computers as a model for human cognition; proposed that human memory was limited in space and duration like computers

- Memory is an integrated system that processes information; Acquire, store, and retrieve information; Components of memory do not act in isolation

Stage forgetting occurs

Any of the 3

Memory is...

...the system of processes involved in retaining, retrieving, and using information about stimuli, images, events, ideas, and skills after the original information is no longer present.

Memory caveats:

- Human memory stages are not as "neat" or independent as in computers

- New learning is dependent on knowledge

- "Successful" acquisition is dependent on how that information is to be used later

Knowledge

Previously learned information

Structural Features

Sensory memory; Short-term memory; Long-term memory

Sensory memory

All or most information that hits sensory receptors. Visual information is called iconic, auditory information is called echoic. We apply attention to information we want in STM.

Short-term memory

Stores small amounts of information for a brief duration. Includes both new and old information; Information just received from sensory memory and information retrieved from long term memory. Maintain activity in STM by rehearsal.

Long-term memory

Virtually limitless storage of memory; unlimited capacity for an unlimited time period. More difficult to retrieve or store information than STM.

Control processes

Active processes that can be controlled by the person

- Rehearsal - Maintenance - reciting/repeating

- Strategies used to make a stimulus more memorable

- Strategies of attention

Rehearsal caveats

Fine to maintain info in your STM; Not particularly effective at committing info to LTM. Thinking deeply works better than repetition

Strategies used to make a stimulus more memorable

- Relational/elaborative rehearsal - Deep thinking - Linking information to existing memories; Levels of processing theory: we can process something "more deeply" which just means really thinking about it's meaning and tying it to things we already know

Neurological significance of Rehearsal vs. Relational/elaborative rehearsal

Brain scans during encoding showed words that people remembered using relational/elaborative rehearsal during recall task were more likely to increase frontal and hippocampal activation. I.e., Deeper thinking--meaning, creation of mental picture--engages higher level thought processes and results in better chance of LTM commitment.

Sensory Memory

Short lived representation of the world (often referring to visual or auditory); Information decays quickly, but it's usually very accurate.

i. Our sensory memory is why we can watch movies and not just see a series of frames/pictures; it looks fluid because our sensory memory is holding each frame long enough for the next to overlap

ii. Without iconic, we would see things as if constantly under a strobe light

Sparkler: persistence of vision

Sensory memory retains the perception of light in memory longer than the actual presence of the light from the sparkler, thus we experience a trail when the sparkler moves.

Incidental vs. Intentional learning

Perform about the same

Sperling's partial report experiment - Key takeaway:

Sensory memory is very accurate, but very short. As Sperling added more time between initial visual stimulus and recall task, (pause before testing sequence) the averages dropped significantly.

Sperling's partial report experiment - Test:

Full Report: Twelve letters flashed, subjects asked to report as many as possible (Average 4.5 of 12 letters)

Partial Report: Tone followed iconic sensory memory which cued participants to report a specific row of letters. Participants reported with higher accuracy (Average 3.3 of 4 letters)

Delayed Partial Report: Same as partial report, except the tone was delayed. The ability for participants to report the letters they saw was reduced significantly.

Working memory

Limited capacity system for temporary storage and manipulation of information for complex tasks such as comprehension, learning, and reasoning. Similar to, but consists of multiple parts and is more sophisticated than original conceptions of STM.

Working memory vs. Short-term memory

- STM is a single component, WM consists of multiple parts

- STM holds information briefly, WM manages processing and manipulation of info occurring during complex cognition

Why does deeper processing help get things into the LTM?

a. Creating connections: multiple retrieval paths can be used to find the info later

b. More connections to things already in LTM = more likely to be able to find/retrieve it

Example: "The hawk swooped down and carried off the chicken." Remembering the last word in a sentence is easier when the sentence is elaborate, including details, because the extra info is helpful in encoding; it creates more connections.

Serial position curve/primacy effect

...illustrates the tendency to recall first and last items in a series best, and middle items worst.

Primacy effect - Information received first is rehearsed longer; more information makes it to more permanent storage. More accurate in recall tasks than information in "middle." Associated with hippocampus (LTM).

Recency effect - Information received last is more recent. Accuracy is attributed to the recency of the stimulus, more accurate than information in the middle of task again.

Associated with perirhinal cortex (WM).

Short term versus long term memory

STM = Small amount of info for a small amount of time

LTM = Unlimited amount of info for unlimited amount of time

i. Key differences - Capacity, duration, ease of access

ii. Cues and hints can help search long term memory

iii. If something leaves STM w/o encoding to LTM, cues won't help retrieval

Working Men. = STM + X,Y & Z

1. Manipulate information and complete tasks

2. STM = single component, WM consists of multiple parts; Working memory is concerned with processing and use of info, not just holding it

Serial/position

Where in a list a word/number/etc was given

Why recency effect?

Working memory - Last things a subject hears are still in short-term system; can be immediately reported without encoding to LTM.

Recency effect experiment:

If people do a task immediately after hearing/reading the list, the recency effect is severely reduced; working memory is utilized elsewhere, which crowds the words out of the list.

Why primacy effect?

LTM - the word at the beginning of a list gets the most rehearsal; these words are more likely to be transferred to the long term store.

Primacy effect experiment:

Slowing word presentation gives subjects more time to rehearse information; primacy effect is increased due to increased time for rehearsal of words at the beginning of the list.

Working memory - key ideas

Critical to day-to-day functioning; used for reading, goal-driven behavior, conversing. Some cognitive abilities can be predicted by WM capacity.

Studying working memory duration

Tested with recall tasks; occupy subjects to prevent rehearsal and test duration and recall capacity. W/o rehearsal recall diminishes. Greater time between stimuli and recall task increases deficit in performance.

Proactive interference

In contrast to original WM decay theories. Information from previous trials interferes with subjects recall capacity. Original ideas about deficient recall task performance was attributed to decay, but later concepts incorporated interference effects posed by multiple trials.

Capacity - Ericcson et. al., 1989

Trained an average college student to remember a large amount of digits by utilizing chunking (achieved 79 digit span with meaningful chunks).

Capacity - Chase and Simon, 1973

Chess board memory test. Chess experts remembered orientations of pieces when they made sense according to game play. When random placement of pieces, both expert and control group performed similarly.

Digit span

How many digits a person can remember; Typically 5 to 8 numbers (Phone numbers); However, didn't take "chunking" into account

Chunking

Chunk is a collection of elements with strong associations to one another, more so than elements of other chunks.

Operation span (reading span)

Measures working memory; replaced digit span. Reading Span test - must remember the last word of increasing length series/sentences.

Working Memory capacity & Intelligence

Measures the ability of working memory to hold information and work with it; the higher someone's score, the higher their score on verbal comprehension, verbal SAT, reasoning and ability, and IQ tests.

Coding

How information is represented/stored in the brain. Physiological and a mental components:

Phys. - How the brain physically represents this, how the neurons fire.

Mental - How stimuli are represented in the mind; an inner voice or an inner image; auditory, visual, and semantic are the three mental representations of info.

Coding - Auditory

The brain's default method of coding even when stimuli is presented visually. The Conrad experiment: Flashed letters subjects reported back what they remembered. People tended to make sound-based mistakes, not visual mistakes (i.e., phonological loop - shown a T report a C rather than an F even though an F looks much more like a T; C sounds more like T)

Coding - Visual

Things w/o easy/clear auditory coding, will be coded visually; can still keep this info in working memory even though the brain doesn't use default auditory coding method

Coding - Semantic

Meaning matters for how information is encoded.

Proactive interference builds over multiple trails, but a dissimilar word can be recalled independently of interfering words because meaning impacts encoding.

Wickens experiment:

Over 4 trials subjects listened to three similar words, and then counted backwards for 15 seconds before recalling their three words. Recall capacity declined with successive trials. The experimental sample subjects were then given a new dissimilar word which they had no trouble remembering because the new word was not affected by proactive interference.

Baddeley's Model:

Originated concept of working memory, non-unitary model. Model has with three components: Central executive; Visuospatial Buffer; Articulatory rehearsal loop.

Central Executive

Attention controller; focus, divide, switch attention.

Controls suppression of irrelevant information; suppresses habitual responses when necessary (you always turn right to walk to the shuttle, but today you're going to Twigg's); plan and set goals.

Visuospatial Buffer/Sketchpad:

The visual and spatial portion of working memory, used for visual imagery, mental rotation, etc.

Articulatory rehearsal loop

Aka Phonological loop (PL)

The inner voice people use to rehearse information in their heads. The phonological loop deals with verbal and auditory info in the WM.

Word Length effect--PL takes longer to rehears long words so they are not as well rehearsed as shorter words.

Word Length effect:

People struggle to recall longer words. This is evidence that we have auditory coding for words; subvocal speech takes longer for longer words. If something takes longer to say in your head, you will not be able to rehearse it as much/you will not remember it as well when asked to recall.

Phonological similarity effect:

We confuse similar SOUNDING letters instead of similar looking letters because we are auditory coding is used for words; we are rehearsing internally using our PL.

Articulatory suppression:

Preventing subvocal speech - May be accomplished by repeating unrelated word out-loud, occupying the phonological loop; forces visual encoding. Word length affect and phonological similarity effect are significantly reduced because AS prevents auditory encoding. AS also reduces memory span.

Brooks experiment Part I:

Subjects memorized a sentence, considering whether words belonged to the category "noun". Two conditions: Phonological--in which answers were verbalized--or visual--in which subjects pointed to the answer. It was easier for people to point to Y or N, rather than say yes or no out loud, because memorizing the sentence was already utilizing the PL. Verbalization of answers further taxed the PL, whereas pointing primarily utilized the VS.

Brooks experiment Part II:

Subjects visualized "F", starting at a top left corner, and had to report whether "F" ended up in an interior or exterior corner. Same two conditions: Verbalizing or pointing at the response. Under the visualization condition, subjects had an easier time answering verbally.

Brooks experiment indicates:

...that the working memory is not unitary- it is set up to process different types of information simultaneously. WM is a system of multiple components; it's hard for the system to deal with tasks/info using the same component (i.e. all visual or all auditory). If the task and the response draw on the same WM component, performance is worse than if the task and the response are distributed between WM components.

Revised Baddeley & Hitch Model:

Adds 4th component, the episodic buffer.

Episodic buffer:

Backup store that communicates with LTM and WM components; has higher capacity and longer duration than either phonological loop or visuospatial sketchpad.

Prefrontal cortex role in working memory:

Responsible for processing incoming visual and auditory info. Remove monkey's prefrontal cortex and they perform at about 50% capacity in delayed-response task.

Funahashi et al. (1989) delayed-response task:

Monkey's were cued to fixate on an X in the center of a computer screen. A blue square appeared and disappeared elsewhere on the screen. Neurons fired in response to the square even after it disappeared. When the fixation X disappeared the monkeys eyes moved to the spot where they remembered the square. During the delay neurons in the prefrontal cortex are the most active.

Maintenance:

Rehearsal/repetition: Works to maintain info in your STM; ineffective for committing info to LTM, even if repeated over and over.

Relational/elaborative rehearsal:

Deep conceptualization: Linking information to existing memories. Deep conceptualization is more effective for committing info to LTM.

Levels of processing theory/depth of processing theory:

Process info "more deeply"--really thinking about it's meaning and tying it to things we already know.

Neurological differences of levels of processing:

Brains imaged while encoding under two conditions: Thinking about how the word sounded, or forming a vivid mental picture of word. Thinking about a word's sound doesn't require deep processing; we can just process the physical attributes of the word. Visualizing the word requires deep processing. Deep processing engages regions of the PFC, rehearsal engages regions in the temporal lobe.

Intentional learning:

Something you try to remember (i.e. studying for a test, trying to remember your grocery list, etc)

Incidental learning:

Something you remember without intent (i.e. what you had for breakfast or who you ran into in the hall earlier)

Intentional vs. Incidental Learning:

Participants given words were told to either rate "how pleasant" words were--not trying to remember, just exposed to it--or just to memorize the list. i. Incidental learners with deep processing and intentional learners performed about the same. Other groups counting letters or finding the "e" in the words didn't perform well at all.

Why does deeper processing help get things into the LTM?

It creates multiple retrieval paths to find info later. More connections to info already in LTM facilitates retrieval.

Testing effect:

Individuals who rehears information in the short term perform better on immediate free-recall tests, but testing without rehearsal resulted in better long-term retention. I.e., testing forces deep processing and commitment to LTM.

Creating cues - cued recall:

Participants that create their own cues outperform those given cues. Those with their own cues recalled 90% of 600 words.

Mnemonics

need to be meaningful to work

Pegwords:

Pre-establish words associated with things to remember; only works when it's a well established practice.

Interactive imagery:

Visualizing the info/situation; works well for most

Organization of info...

...Key to creating connections in to-be-remembered material. has to be in place during encoding; it needs to be a preexisting structure for use mnemonics, pegboards, imagery, etc., structure can't be imposed after information is coded.

Meaning & memory connections (elaborative encoding)

Depth of processing (paying close attention to meaning) promotes recall by facilitating later retrieval. Elaborate sentences aid memory. Ex: "The great bird swooped down and carried off the struggling chicken." vs "She cooked the chicken." Complex sentences → better memory

Results in richer retrieval paths.

Long term memory:

Archived information about learned knowledge and experiences. Works closely with WM. Appears to be unlimited in capacity, but recent memories tend to be more detailed.

LTM or WM? Same thing?

Not the same. We can lose our ability to form new LTMs and maintain WM capacity, or vice versa.

Clive wearing, H.M., Mr G:

Functioning STM, no LTM

Anterograde amnesia:

no new, explicit memories

Explicit vs. Implicit memory:

Things were are consciously aware of vs. unconscious memory; Episodic memory (things/events about your life); Semantic memory (general, factual knowledge that isn't about our own life).

vs.

Processes going on outside our conscious awareness unconscious memory; "Muscle memory"; things that the body remembers, like riding a bike, also priming, classical conditioning, and nonassociative learning

Matching Learning & Test Condition

Matching the context at the moment of retrieval with the conditions during encoding can help retrieve information from memory.

Context-dependent learning:

Learning dependent on the state one is in during acquisition. Experiment - some learned a list of words under water, some learn on land. Subjects were then tested either under water or on land. Matching test with encoding conditions (i.e. test and learning on land) improved performance.

Encoding specificity (the divers)

I.e., matching test with encoding conditions (i.e. test and learning on land) improved performance.

Study effectively:

Elaborate; Generate and test; Organize; Take breaks; Match learning and testing conditions; Avoid "illusions of learning"

Clive Wearing...

suffers from anterograde amnesia (meaning he can't create new memories) as well as retrograde amnesia (meaning he's lost many of his memories).

anterograde amnesia

unable to create new memories

retrograde amnesia

loss of past memories

Elaborate;

Highlighting is insufficient; more elaborate memories and traces ease recall (imagery, stories, etc).

Generate and test;

Testing forces deep thinking; teaching info to someone (generating it) is good for commitment to memory. Generating test questions forces complex conceptualization.

Organize;

Helps reduce the load on your memory

Take breaks;

60 minutes of studying is better as 3 20 minute chunks rather than 1 60 minute chunk. It is hard to maintain attention for long periods and returning to material after a study break, allows for progress assessment.

Match learning and testing conditions;

Match learning and testing conditions

Avoid "illusions of learning"

Familiarity does not mean comprehension

Context reinstatement:

Context is encoded with info, so it affects retrieval. Memory trace is more likely to match if context/surroundings match those experienced during encoded.

Consolidation:

Sleeping after you study improves consolidation in the memory; creates stronger tracing.

Memory network:

One of many models of how the long-term memory system represents and retrieves explicit memory info.

Spreading activation:

Activating one node/concept, and activity spreads to other connected nodes (associated concepts or ideas). The strength of connections between nodes depends on personal experience with related concepts. Spreading activation helps explain why hints or cues help us remember. Mutual activation of two networks can be enough to exceed the node's threshold to fire.