Chapter 7 – Human Memory
Semantic memory: memory for general information
Episodic memory: memory for personal events, allows us to “time travel” (Tulving’s K.C. did not
Encoding: getting info in, forming a memory code, requires attention (entering data in a
Storage: maintaining encoded info in memory over time (saving a file)
Retrieval: recovering info from memory stores (opening up a file)
Active encoding is a crucial process in memory and requires attention next in line effect,
don’t remember what the person before you said cuz didn’t pay attention or encode the info
Attention involves focusing awareness on a narrowed range of stimuli or events
Selective attention is crucial for a life without chaos
Attention is filter that screens out most potential stimuli but allows selected few to pass through
Debate whether the stimuli are screened out early (during sensory input), or late (after the brain
has processed the meaning or significance of the input)
Cocktail party phenomenon: talking in a party with someone, but as soon as you hear your name
in another conversation you notice it without having been paying attention to that convo late
selection based on meaning of input
Late, early and intermediate selection have all been supported by scientific evidence some
suggest that the location of the filter is flexible and depends on our “cognitive load”
When doing high-load tasks, consumed most of attentional capacity, selection tends to occur
Low-load tasks, attentional capacity left over to process meaning of distraction later selection
Hard to focus attention on two or more inputs at once, devided attention between memory
encoding and some other task large reductions in memory performance
Situlus | recognition of meaning | response slection response
Models of selection, early selecting filter before recognition of meaning, late selection is after
the recognition of meaning, we may be able to move the filter back and forth between these
Human brain can only effectively handle one attention-consuming task at a time, when
multitasking just switching attention back and forth among tasks, not processing them
Some info that we want to intentionally remember encoded as a result of effortful processing
Some info acquired more automatically, e.g. frequency of words
Verbal elaboration strategy to remember found to be associated with activity in a network of
regions including prefrontal cortex
Visual imagery strategy showed brain activity in the extrastriate region next to the primary
Different rates of forgetting occur because some methods of encoding create more durable
memory codes than others Dealing with verbal info, people engage in three progressively deeper levels of processing
resulting in longer lasting memory:
Structural encoding: shallow processing, emphasizes physical structure of the stimulus (eg.
Phonemic encoding: how a word sounds like, naming or saying (out lout or silently) the words
Semantic encoding: meaning of verbal input, thinking about objects and actions the words
Time required for processing is not a good indicator of the depth or level of processing,
impossible to design a task in which structural encoding takes longer than semantic encoding,
processing time is not a reliable index of depth
Elaboration, visual imagery, and self-referent coding can enrich the encoding process and
Elaboration is linking a stimulus to other info at the time of encoding
Thinking of examples that illustrate an idea effect memory dramatically, self-generated example
would probably be more valuable in enhancing memory.
Additional association created by elaboration helps people remember info
Visual Imagery: creation of visual images to represent the words to be remembered, also
Easier to form images of concrete objects than of abstract concepts, ease of image formation
affects memory, high imagery words are easier to remember
Imagery facilitates memory because it provides a second kind of memory code.
Dual coding theory: memory is enhanced by forming semantic and visual codes, since either can
lead to recall.
Self-referent encoding: making info personally meaningful, deciding how or whether info is
Self-referent encoding seems to enhance recall by promoting additional elaboration and better
organization of info
Self-referent encoding used as research tool to show that depressed people have more negative
content as part of their self-prototype
Aristotle and Plato compared memory to a black of wax that differed in size and hardness for
Information-processing theories (analogy to info storage by computers_ subdivided memory
into three separate memory stores: Two short temporary buffers (sensory store and short-term
store), and long-term store
Sensory memory preserves info in its original sensory form for a brief time (a fraction of a
second) giving you additional time to try to recognize the stimuli
Sensation of a visual pattern, sound or touch to linger for a brief moment after the sensory
stimulation is over, eg. Afterimages
Memory trace in the visual sensory store decays in about ¼ of a second, auditory traces also last
less than a second Short-term Memory: limited-capacity store that can maintain unrehearsed info for up to about 20
Rehearsal: process of repetitively verbalizing or thinking about the information alloews maintain
info in STM indefinitely
Maintenance Rehearsal: simply maintaining info In consciousness
Elaborative rehearsal: increasing the probability that you will retain the info in the future (eg.
Meaning of words)
Durability of Storage: loss of info from short-term memory due to time-related decay of memory
traces but also due to interference from competing material
STM has limited capacity (about 7 unrelated items), constrains people’s ability to perform tasks
in which they need to mentally juggle various pieces of info
Can increase capacity of STM by grouping familiar stimuli as a single unit called chunks
Draw info from longterm memory to evaluate and understand info in STM
Experts in a field chunk infor differently and more effectively better able to remember things
related to that field e.g. chessmasters
STM as Working Memory:
STM not limited to phonemic encoding, decay is not the only process responsible for the loss of
Working memory (by Baddeley) has four components:
Phonological loop: evolved to facilitate the acquisition of language, eg. when you recite a phone
Visuospatial sketchpad: permits people to temporarily hold and manipulate visual images e.g.
arranging furniture in your mind
Central executives system: controls deployment of attention, switching the focus of attenintion
and dividing it as needed, also coordinates actions of the other modules
Episodic Buffer: temporary, limited-capacity store that allows the various components of
working memory to integrate info, also works as an interface between working memory an long-
Separate components of working memory work individually.
Variations in working memory capacity correlate with measures of high-level cognitive abilities
Working memory capacity plays a fundamental role in complex cognitive processes
Long-Term Memory: unlimited capacity store that can hold info over lengthy periods of time
One theory is that LTM can store info permanently, only forgot cuz you cant retrieve info,
however no convincing evidence to support it.
Long lasting memories triggered by ESB (evidence for permanent theory) but the memories
where hallucinations and inaccurate
Flashbulb memories (unusually vivid and detailed recollection of momentous events) are also
evidence for permanent storage (eg. Where were you when Sidney Crosby scored the goal) but they are not as accurate or special as once believed, become less detailed and complete with
time, often inaccurate.
Originally believed that STM depended on phonemic encoding and LTM on emantic encoding,
STM info loss was due to time-related decay while LTM info loss due to interference. These
distinction have been undermined by research evidence: semantic encoding and interference
effects found in STM
Other perspectives on STM and LTM is that STM is a tiny and constantly chanign portion of LTM,
another view is that there is only one single, generic momory store governed by one set of rules
Clustering and Conectual Hierarchies:
Spontaneously organize factual info into categories for storage in memory, clustering or
remembering similar or related items in groups.
When possible this info is organized into conceptual hierarchies: multilevel classification system
based on common properties among items
Schemas (prof’s office example): organized cluseter of knowledge about aparticular object or event
abstracted from previous experience with the object or event
People are more likely to remember things that are consisten with their schemas than things
that are not.
People sometimes exhibit better recall of things that violate their schema-based expectation
(might attract more attention and deeper processing)
Relational schemas: representations of typical events surround interpersonal interactions that
represent regularities in your interpersonal experience, affect the way you process info about
other s and yourself.
Semantic Networks: consist of nodes representing concepts, joined together by pathways that link
related concepts (e.g. thinking about butter leads to bread, thoughts naturally go to related words,
called spreading activation)
Ovals are notes, works inside the ovals are interlinked concepts, the lines connecting the nodes
are the pathways, the shorter the pathway the stronger the association
Connectionist Networks and Parallel Distributed Processing Models: assume that cognitive
processes depend on patterns of activation in highly interconnected computational networks that
resemble neural networks.
Consists of nodes or computing units, operate like neurons, be inactive or may send either
excitatory or inhibitory signals to other units.
Specific memories correspond to particular patterns of activation in these networks Different from semantic networks since nodes don’t represent specific concepts or pieces of
knowledge, but a piece of knowledge is represented by a particular pattern of activation across
an entire network, info lies in strengths of the connections.
A node’s level of activation reflects the weighted balance of excitatory and inhibitory inputs
from many other units.
Inspired by how neural networks appear to handle info; parallel distributed processing
Tip-of-the-tongue phenomenon – temporary inablility to remember soemtihing you know,
accompanied by a feeling that it’s just out of reach.
Retrieval cues: stimuli that help gain access to memories
Encoding specificity principle: memory for info would be better when conditions during
encoding and retrieval were similar.
Cues used at retrieval will facilitate recall iff the info about them and about their relation to the
to-be-remembered words is stored at the same time as the to-be-remembered info
Context Cues, eg. Going back to an old home, enhance memory recalling, used on eyewitnesses
Hypnosis often increases subjects’ tendencies to report incorrect info.
Cues also related to state and mood-dependent effects too, eg. Better recall of info gathered
when intoxicated while you are intoxicated.
State dependent memory effects are more easily obsereved for free-recall conditions than for
more readily observed for recognition or cued-recall where retrieval cues are more abundant.
Negative emotions may increase one’s susceptibility to false memories, misinformation effect.
When pull info from LTM, memories are sketchy reconstructions of the past that may be
distorted and my include details that did not actually occur (eg. War of ghost example)
Part of memories are the details of the event and part of it is the reconstruction of the event
based on their schemas
Misinformation effect (partly due to to unreli