Transcript Chapter 5 – Cognitive Engineering

Cognition, Memory, and
Attention
ITM 734 Fall 2006
Dr. Cindy Corritore
Through all of this ….
 limited cognitive resources



analogy
flawed plans (heuristics)
simulations (cognitive/mental models)
 goal – to minimize complexity through
improved fit (between user, computer, and
task)
Copyright 2006 Corritore
2 of 52
memory types
 sensory memory
 short-term memory
 long-term memory
Copyright 2006 Corritore
3 of 52
sensory memory/store
(multi-store theory)
 buffers for incoming data via senses
 different one for each sense
 types
 iconic store – visual store; fades rapidly – can operate
on this store
 echonic store – auditory store  short-lived and space-constrained
 persistence (fireworks in vision after the fact)
 some processing even if not attended
 attention brings it into STM
 cocktail party phenomenae
Copyright 2006 Corritore
4 of 52
STM characteristics
 quick access and quick decay (volatile)
 limited in size
 chunking (experts vs. novices) - phone number
 402-111-5555
 forgetting
 time decay?
 interference with new items? (eg. similarity)
 attention moves off item?
Copyright 2006 Corritore
5 of 52
STM
 gateway to sensory and LTM?

no – conversation goes directly to LTM
 role of rehearsal exaggerated (moving from
STM to LTM)

lots in LTM that is not rehearsed (eg. snapshot
of a birthday celebration)
Copyright 2006 Corritore
6 of 52
STM characteristics
 recency - last few items in list recalled better
than middle - holding most recent items in
STM


negate with interference?
visual and auditory channel - no interference
if different channel
 primacy - first few items in list recalled better
than middle (more rehearsal)
Copyright 2006 Corritore
7 of 52
LTM characteristics
 Slow but variable access speed
 Permanent (little decay)
 Infinite capacity
Copyright 2006 Corritore
8 of 52
LTM characteristics
 Retrieval depends on ….
 recency
 expectations
 similarity of information
 connectedness
 rehearsal
 richness & nature of processing at learning
 level or depth or processing (shallow vs deep
perceptual analysis)
 distinctiveness of processing
 amount of processing
 elaborate far better
Copyright 2006 Corritore
9 of 52
Richness
 paragraph – listen and remember …..
Copyright 2006 Corritore
10 of 52
Types of LTM
 Explicit and Implicit


conscious recollection, top-down retrieval
from multiple systems with massive integration
(E)
unconscious recollection, bottom-up from
single system (I) – more automatic
Copyright 2006 Corritore
11 of 52
Types of LTM
 Episodic and Semantic



self-awareness component, things that
happen to you, complex (E)
stuff we know, knowledge about the world,
relationships, implicit - dictionary, thesaurus
likely stored the same way
Copyright 2006 Corritore
12 of 52
Types of LTM
 Declarative and Procedural


knowing that, explicit primarily, relationships,
integration of information (D) – knowing things
and their relationships
knowing how, mostly implicit, not relational –
how to do things
Copyright 2006 Corritore
13 of 52
Memory structures for stories,
events …
 Schema - framework that includes frames & scripts
 become chunks for expanding memory
 basis for expectations
 Bartlett’s Schema Theory
 framework for stories that affects comprehension
 told American Indian stories, then recall  readjusted story elements and themes to fit their model
 Chunking in experts
 Helps make it easier to recall, group information
 Experts have great, robust schema and chunks
Copyright 2006 Corritore
14 of 52
Everyday memory
 little studied
 appears to have a lot of variability


eyewitness memory
flashbulb memory
Copyright 2006 Corritore
15 of 52
Eye-witness memory
 Effects




post-event memory - questioning right
after the fact can distort (retroactive
interference)
verbal overshadowing - talking about it
right after happens over-writes visual
memory
memory in the world sketchy (Normal)
confirmation bias - affected by what you
expect
Copyright 2006 Corritore
16 of 52
Flash bulb memory
 what were you doing when heard about 911
disaster?
 Richness …..
Copyright 2006 Corritore
17 of 52
LTM processes
 Storage

rehearsal
 Retrieval



Forgetting
Recognition vs recall
Frequency and recency effects
Copyright 2006 Corritore
18 of 52
Storage- Rehearsal
 Memorization involves storing the information and





one or more access paths
Good memories are rich semantic networks with
many (unique) access paths
Learning is aided by meaningfulness, structure,
familiarity and concreteness
Active memorizing requires effort, motivation
Passive memorizing - unpredictable, often episodic,
context sensitive
Similar items interfere if they are not separated
during memorizing - learning transfer effects - old
interfere with new; new overwrite old
Copyright 2006 Corritore
19 of 52
Facilitating Memorization
 Structure information to help chunking - use
categories, ordering, associations
 Encourage reasoning during memorizing active memory
 Help access by multiple pathways memorizing tricks e.g. keywords, cognitive
aids, mnemonics, link to image memory
(rooms)
 Make associations clear and keep them
consistent
Copyright 2006 Corritore
20 of 52
Facilitating Memorization
 Make separate and recognizable contexts for
recall - important for script / skill memory
 Increase depth of encoding




Richness
Visualization
Uniqueness
Interaction
 Recognition
Copyright 2006 Corritore
21 of 52
Facilitating Memorization:
Mnemonics
 cognitive mnemonics


ABC’s with tune
on old olympus mountain tops a finn and
german viewed some hops (12 crainial nerves:
OOOMTAFAGVSH)

seems to be more to remember?
Copyright 2006 Corritore
22 of 52
Facilitating Memorization:
Mnemonics
 check out:
http://human-factors.arc.nasa.gov/cognition/tutorials/index.html
 mnemonic for Norman principles:
visibility, feedback, cognitive/conceptual model, affordance,
mapping
My Fat Cat Ate Veggies
Copyright 2006 Corritore
23 of 52
Retrieval - Theories of forgetting
 repression (Freud) - bad experience
 interference (proactive or retroactive)
 previous learning/memories interrupt
 espc if similar stimuli – belong to same category
 eye-witness and post-incident questioning

doesn’t explain how it works
 cue-dependant
 forget because info not there anymore or
*can’t access it
 encoding specificity principle (cue-dependant)
 retrieval a func. of overlap between
information present at retrieval and info stored
in memory
 includes contextual info

Recognition dependant on internal cues only
(not external context)Copyright 2006 Corritore
24 of 52
Recall vs. recognition
 Knowledge in the World Theory is GUI’s - Alan Kay
developed in 1960’s





Steve Jobs in late 1970’s from Xerox Parc
keep knowledge in world to supplement head knowledge
recall vs. recognition
remember just enough detail to get by
 exceptions rather then norms
experts not expert in knowledge in the head as much as
expert in how to locate needed knowledge in the world
(Norman Ch 2)
Copyright 2006 Corritore
25 of 52
Design implications
 Reduce cognitive load!!!
 Type of user

novice, expert, intermittent user
Copyright 2006 Corritore
26 of 52
Design implications
 Mental models natural extensions of schema -
support schemas
 metaphors - desktop/office
 match system information structure with
familiar memory structures so user can use
their schema
Copyright 2006 Corritore
27 of 52
Design implications
 Design interfaces that help users ‘grow’ good
mental models
meaningful and familiar command names (eg. from
task world)
 balance this with existing conceptual models of item
names (ie. cut, copy)
 Incorporate closure (finish) on tasks
 helps build mental model
 helps identify chunks for memory when become an
expert
 Consistency - to build mental model; don’t have to
remember as much

Copyright 2006 Corritore
28 of 52
Design implications
 Rich encoding 


multimedia
interaction
context?
 May just be to ‘remember your site’


http://www.pulse3d.com/pulse/
http://www.jordans.com/roomplanner.asp
Copyright 2006 Corritore
29 of 52
Design implications
 Focus on recognition rather than recall



interface contains prompts/information
studies on computer experts found they don’t
have better recall, but high recognition of what
is and isn’t available on interface and where to
find it (mental maps)
GUI’s combination of recognition (menu’s)
and recall (quick keys)
Copyright 2006 Corritore
30 of 52
Design implications
 Place the burden of remembering on the
machine, not the human
 Don’t require user memory (eg. between
screens)
 Don’t have computer ask for info it can derive
Copyright 2006 Corritore
31 of 52
Design implications
 Design minor messages, alerts, warning to be
minimally disruptive

prevent user from forgetting data stored in
short term memory
Copyright 2006 Corritore
32 of 52
Evaluate these
 http://happydeluxe.com/
 http://www.google.com vs http://www.yahoo.com
 http://www.northcantonmedical.org/
 http://www.enchantedharp.com/
Copyright 2006 Corritore
33 of 52
Attention
 Humans can focus mental resources on a
single event/object



helps to simplify environmental input (filter)
works with perception - perceive what
attending to
can divide attention (multiprocessing, not
parallel)

problem - distraction on second task, don’t return
to first task in right place.
 often use world reminders to hold place in first task
(post-it note)
Copyright 2006 Corritore
34 of 52
Attention
 examples


driving a car -must attend to some stimuli,
ignore others
listening to this lecture - attend to slides and
words, ignore other students, physical plant
noises
Copyright 2006 Corritore
35 of 52
Divided attention
 doing two things at once
 affected by



task similarity – similar how?
practice (experience) - automaticity
task difficulty – require more resources
than are available?
 what happens: interference
Copyright 2006 Corritore
36 of 52
Success in time sharing attention
 four mechanisms account for how well we
divide our attention
1. automaticity and resources
2. resource allocation and switching
3. structural factors
4. confusion and similarity
Copyright 2006 Corritore
37 of 52
1. automaticity and resources
 Automatic vs. Controlled : perform task without
thinking about it or require attention, conscious
control. Happens over time. Controlled – do
something directed by thought.
 Automatic:


good as fast, doesn’t interfere with other tasks
(need minimal attention), unconscious
bad - unavailable to conscious level, hard to
change (driving a shift), can interfere with other
automatic processes, harder to unlearn

do experiment: Stroup
Copyright Effect
2006 Corritore (http://www.apa.org/science/stroop.html)
38 of 52
1. automaticity and resources
 automatic processing can time-share efficiently

doesn't require a lot of cognitive resources


eg. walking
factor: effort and difficulty of additional tasks
 if task difficult, requires more resources


if have dual tasks, performance will decrease
since resources are being shared
automatic tends to reduce the difficulty
Copyright 2006 Corritore
39 of 52
1. automaticity and resources
 can only increase performance so much

level equal to ‘full’ resource use on a task,
performance data limited (no further benefit
from adding more resources)


perfect example: no matter how hard I try (invest
resources & effort), I won't improve my
understanding of a discussion in French beyond a
rudimentary level.
also called resource-limited
Copyright 2006 Corritore
40 of 52
1. automaticity and resources
 bottom line



increase effort into a task, improve
performance to point if resource limited
increase difficulty of task decreases
performance unless add resources
in dual tasks, if increase resources for
one task, will decrease resources for
second task and subsequent
performance

depends on automaticity
Copyright 2006 Corritore
41 of 52
2. resource allocation and
switching
 result of two + tasks co-occuring
 now look at how you can allocate and switch
attention between tasks

we don't have elaborate schemes to optimize
resource allocation


can improve time sharing with these strategies
totally depends on the individual
 can train how to control attention
Copyright 2006 Corritore
42 of 52
2. resource allocation and
switching
 optimal allocation schedule vs. actual based on
task importance and other factors
 factors



switch cost (so tend to stay with same task even
if low priority)
cognitive distance of tasks - if close, more
confusion when switch (so more costly)
faster switch if salient reminders available about
task (eg. you can see it vs. just remembering)
Copyright 2006 Corritore
43 of 52
3. structural factors
 perceptual resources required, brain structures
used, info processing required
 Bottleneck Theory- use same resources, get a
bottleneck that shared tasks must wait for
 bottom line

amt. of interference between two tasks depend
on degree to which each requires same
resources (shared levels on these three
dimensions)
Copyright 2006 Corritore
44 of 52
4. confusion and similarity
 confusion: increasing the similarity of
processing material decreases efficiency (too
similar)

eg. mental math and spelling, Stroup effect

semantic value of word interferes with ability to
report ink color
 what happens: responses for one task activated
and interfere with second task
 eg. two verbal tasks, one requiring working
memory and the other active processing (eg.
comprehension)
Copyright 2006 Corritore
45 of 52
Visual attention theories
 spotlight vs. zoom lens

both correct in part, likely zoom is more
appropriate (zoom focus in on what’s imp)
 how attention works


overall gestalt (salient features), focus down
on objects and components
affected by experience (bananas yellow)
Copyright 2006 Corritore
46 of 52
Designing for attention
 examine which configurations minimize task
interference

voice recognition software - may interfere if
user has to perform other verbal activities

best with spatial activities
 avoid imposing two tasks using similar
materials (confusion)

entering digits while others speaking digits
Copyright 2006 Corritore
47 of 52
Designing for attention
 what about background music?


requires spatial perception
decreased performance with lyrics and word
processing
 examine mental workload
Copyright 2006 Corritore
48 of 52
Designing for attention
 Ways to focus user attention

structure information

group like things
 physically, with fonts, with color, spacing, lines, etc.




use same spot for same types of information
to help with distractions: system should inform you
where you were in task when left
 let user know position in state space
avoid unnecessary information display (KISS)
make things easy to use/move thru (so user not
focused on mechanics of how to use system)
Copyright 2006 Corritore
49 of 52
Designing for attention
 taking advantage of automatic processing:
quick keys across systems

standards (like Windows 95 ^c, ^v, ^x) become
automatic

problem - appears unrelated to task to most people
 avoid automaticity by interrupting process (eg. put
a window up in middle of keystroke sequence)


good for deleting
urgent info in prominent area; less urgent to
specific area(s)
Copyright 2006 Corritore
50 of 52
Errors
 how design for them? will occur

save the user …..
 Why?


we are satisficing, not optimizing
always following a plan (heuristics), but most of
our plans are flawed
Copyright 2006 Corritore
51 of 52
Action slips
 activate wrong schema
 Norman discusses 6 types of slips - a result of
different kinds of automaticity errors






capture errors – frequently done takes over (same
beginning)
description errors – intended action fits several
possibilities – pick wrong one
data-driven – data interrupts automatic behavior and
get wrong behavior
associative activation – trigger activates wrong action
loss-of-activation – forget why doing something
mode – more than once state possible
 slips vs. mistakes (choose inappropriate goals)
Copyright 2006 Corritore
52 of 52
Errors ala Te’eni et. al.
Copyright 2006 Corritore
53 of 52
Designing for errors
 Schneideman’s guidelines good
 specificity - what exactly is the problem?
 constructive guidance - how can user fix/deal with
problem?
 positive tone - vs. illegal, aborted, fatal…
 user-centered style - how phrase suggestion
 appropriate physical format - mixed case,
placement on screen?
 Guidelines from Microsoft
http://msdn.microsoft.com/library/default.asp?url=/
library/enus/debug/base/error_message_guidelines.asp
Copyright 2006 Corritore
54 of 52