Standard view — Memory has three main modules:

1. Sensory store: Maintains sensory information for extremely brief periods of time.
2. Short-term memory: Contains the information that we are actually using.
3. Long-term memory: "Permanent" memory.

Iconic memory: Maintains information presented visually for extremely brief (~100ms) periods of time.

Sperling (1960) demonstrated the existence, capacity, and duration of iconic memory.
 
 

Whole-report procedure: Subjects are asked to report every letter they can remember.

Partial-report procedure: Subjects are asked to report information from only one line of the display, but they don’t know ahead of time which line.

(See Sternberg, Fig. 5.3)

This memory holds the information we actually use in our cognitive processes.

Miller’s 7 ± 2 model of STM capacity.

Because STM is seen as the store for the information we are actually using, it is often rethought of as being working memory, to highlight it’s active role.

Baddeley’s model:

Three components —

1. Visuo-spatial scratchpad: Maintains and manipulates visual images in working memory.
2. Phonological loop: Maintains words for remembering and rehearsal.
3. Central Executive: Coordinates the transfer of information between working memory and long-term memory, and controls response preparation.

Long-term memory contains all the "permanent" things you need to know.

Three main pieces of LTM:

1. Declarative (semantic) memory: Memory for facts, such as "Pittsburgh is in Pennsylvania."
2. Episodic memory: Memory for particular events.
3. Procedural memory: Memory for how to do things. Your automatized processes are stored here.

Two different forgetting functions for STM information vs. LTM information.
 

Are the functions really that different, though, or are they just a different scale?

The key idea is this: The more "deeply" you process a piece of information, the more strongly the information will be encoded in memory.

Example: (See Sternberg, Fig. 5.2)

Self-reference effect: Subjects are more likely to remember information that was related to themselves somehow.
 
 

Elaborative processing: The more a subject elaborates a piece of information, the more likely it will be remembered.

Example: Bobrow & Bower (1969) —

Subjects were to remember simple subject-verb-object sentences, i.e. "Bob kissed Mary."

When subjects were given sentences to study that were generated by the experimenter, they were able to accurately recall the object noun when prompted with the subject noun 29% of the time.

When subjects had to generate their own sentences by providing the verb to connect the subject and object nouns, they could recall 58% of the object nouns.

1. Circularly defined: Something that’s better remember must be more deeply encoded, and if something is more deeply encoded it will be better remembered. How can you separate different levels?

2. Context effects: We better remember something if the context of retrieval matches the context of encoding.

Example: When retrieving words based on whether they rhyme with the prompt words, we have better retrieval if we studied rhyming words, regardless of how "deeply" we processed them.

Retrograde amnesia: Loss of long-term memories prior to whatever trauma caused the amnesia.

Anterograde amnesia: Loss of memory after the trauma.

Example: H.M.

Infantile amnesia: We don’t remember things from when we were very young.
 

There are also various deficits of the specific memory systems (semantic memory, episodic memory, etc.) mentioned above.

There is evidence that some of our memory processes are implicit, i.e. they take place without our conscious awareness.

Example: Priming (discussed during attention)

Evidence from anterograde amnesia patients (such as H.M.) that, while explicit memory is severely impaired, subjects are still able to learn tasks, such as drawing a maze in a mirror, without recalling practicing those tasks.