Norman Contributions to Things That Make Us Smart.
There is only so much we can remember, only so much we can learn
Miller, approximately seven items
1) learners try to link new information to what they already know;
(2) knowledge is stored in networked clusters and organized into
Schemata (isolated pieces of information are not recognized as meaningful
And are soon forgotten);
(3) learners actively construct meaning by trying to find regularity
And order in what they see, read, or hear.
Tools
Experiential cognition/learning
Reflective cognition/learning
Technology will affect how and what we learn
Chapter 1
The experiential mode of cognition leads to a state in which we perceive and react
to the events
around us, efficiently and effortlessly. It is the basis of expert behavior. It's essential for driving,
for
playing a musical instrument.
The reflective mode of cognition is that of comparison and contrast, thought,
and decision-
making. This is the mode that leads to new ideas and novel responses. We can clearly see the
difference when we drive on auto-pilot and wind up taking the wrong exit!
A key point that goes throughout the book is "don't let the technology use
you". Any technology
that attempts to make you think like a machine, because it's easier to build a machine than to
understand how the human mind really functions, is going to have pros and cons - you can never
get rid of the cons. (see McKenna on evaluating interactive multimedia learning environments.)
Chapter 2
There are three kinds of learning:
Accretion (gathering facts),
Tuning (practicing skills) and
Restructuring (forming the proper conceptual structures).
The first 2 are experiential modes; restructuring is a reflective mode. The experiential
mode of
performance is one of perceptual processing (pattern-driven or event-driven). This is bottom- up, like
sitcog. In contrast, the reflective mode is conceptually-driven, top-down processing, like
computation or logical deduction.
Remember - our minds function better in an analog fashion, seeing patterns emerge;
whereas
machines have a tough time with this. Machines are quick in computation, but people have a tough
time with this. (Partly, this is because some of our mental processing is pre-attentive - we notice
patterns almost automatically.)
Chapter 3
Tools of thought - cognitive artifacts - are external aids (in the environment,
not in our minds) that
enhance cognitive abilities. A good external representation (representing perceptions, experiences,
and thoughts in some other medium other than that in which they have occurred) captures the
essential elements of the event, deliberately leaving out the rest (irrelevant details are abstracted
away).
A representational system has two ingredients: see Nesher, Microworlds in mathematical
education
Knowledge component
That which is to be represented (like addition & subtraction) and
Exemplification component
A set of symbols, each standing for something in the represented world (like
Cuisinaire rods).
Moreover, we can have meta-representations - representations of representations
- like SU3,
where a triangle of elements (like bowling pins) represents the graph of energy vs. strangeness (the
representation) of a set of elementary particles (the "real world").
The most appropriate representational format depends on the task - no format can
be correct for
every purpose. And, since the user of an information display must (1) find the relevant information,
and (2) compute the desired conclusion, the best representation for a given task is one that
transforms the task from reflection (top down, logical processing or computing) to experiencing
(seeing the pattern emerge). Hence, we use graphs rather than tables to represent trends; and
when we are counting things, it's easier to use tally marks than Arabic symbols. And this is why
group theory is so powerful in particle physics.
Chapter 4
Realizing the limits of working memory, we know that a person plus an external
representation
(artifact) is smarter than the person alone. There are two types of representations:
Surface representations (everything is physically visible, like machinery) and
Internal representations (using arbitrary or abstract relationships to relate
the indicators or
displays to the state of the system).
An abacus is a surface representation of math. MPC1 uses numbers and machine code
to
display internal representations for intermediate results of a computer program; these intermediate
results are really the results of electronic signals going through "and" "or" and
"nor" gates.
Tasks may appear the same to a computer that uses an algorithm, but to a person,
they may
appear very different. Why? If people can make use of physical structures in the environment,
there's less information they have to store in working memory, and the more cognitive power is
freed up. So, if you are timing something very accurately, use a digital stopwatch; but if a cop is
following you, you want to see if you are over or under the "55" red line.
Different technologies have different affordances. Those that rely on space have
very different
affordances from those that rely on time. That's why animated algorithms are much easier to
understand than capturing sequential states of the solution and drawing them in order of
occurrence in a textbook.
Chapter 5
When we try to apply the rules of hard science to the human mind it doesn't work,
because
modern science is based on reductionism. Prigogine says the same thing as Norman - you can't
build up a complex system from simple blocks, just because you know how those simple blocks
work. Remember - humans are social creatures, and a lot of learning is social.
We are different from machines because machines find computation and memory for
vast
databases easy, but they can't deal with deception, the appreciation of beauty, music, etc.,
because these require the ability to represent knowledge and metaknowledge, to form
representations, to compare resentations with each other, and to form causal explanations of
events.
We are different from animals because we are capable of true, social, cooperative
behavior (and
can turn it off when it's no longer appropriate, not like bees), we can teach (break complex
procedures into steps and model how to do each step properly, not just learn through imitation),
and we can make and use complex tools (including tools to make other tools, and cognitive
artifacts).
Human intelligence has evolved from
Episodic memory (remembering events that you've experienced),
Mimesis (the ability to act out or mime intentions, desires, and wants, like
pointing to a watch),
Mythic (communicating rich concepts and throughts, stories, and myths that provide
explanations for the events of life), and
External representation (cf. Allen & Otto - expand our abilities beyond what
our biological
heritage alone makes possible, through writing, external representations, and other tools that use
the affordances of the environment to overcome the limitations of our brainpower).
Self-awareness, too, is characteristically human, as is the awareness of other
people's states of
mind (hence social graces, cooperative behavior, and norms of social interaction).
We prefer to work in an experiential mode (including preattentive processing,
pattern recognition,
recognizing the whole from a part - see Fleming & Levie). However, this often makes us jump to
conclusions. We may talk a lot about using logic to make decisions, but often major decisions are
based on stories (think about Reagan's awful campaign speech about riding along the California
coast!) Why? Because logic is a branch of mathematics, not a branch of natural thought. It's an
abstraction, not concrete. Moreover, logic can only handle aspects of a situation that are thought to
be important - not the whole scene, so it can oversimplify to the extreme (and thus make crucial
errors because the model doesn't always fit the real world, like in computer modeling of radar
paths, or a true experiment in educational research). Logical analysis only applies to what can be
measured - how can we measure beauty and morality? Stories carry context; logic tries to
abstract, to generalize, to remove subjectivity. They are very different!
Errors
Slips and mistakes. We slip when the action performed is not the one intended
(dropping a
dish on the new tile floor). We make a mistake when the intended action is wrong (we run a red
light). Why do we err? Because we remember the substance and meaning of events, not the
details; and because we can only keep our mind on one conscious task at a time. We have a short
attention span. We are sensitive to changing events, not to continual, sustained events. We also
match present patterns to similar events that have occurred in the past. And we have tunnel vision.
Once your mind is set, it's difficult to change it, even in the face of contradictory evidence. In other
words, we can quickly come up with a good explanation for why something happened and how to
fix it (which a machine finds hard to do), but when this property gets sidetracked by the wrong
explanation, it's very difficult to get it redirected - like misdiagnosing an illness. That's why it's
best
to get an impartial observer (a "third eye") when you're getting stuck or things aren't working
out as
planned. I think that's the role of the skeptic in CSILE.
How to deal with this? We are good at creating mental models (unlike machines),
not at highly
accurate repetitious tasks (like machines). Don't require tasks that require using memory for
details, or devoting long periods of attention to unchanging situations. Have the task make sense.
And provide informative feedback to make sure our explanations are in sync with the actual
situation.
Chapter 6
Norman uses the example of an airplane cockpit to discuss how old, "obvious"
technologies
actually help with informal communication among a team that's engaging in a common task (e.g.,
flying a plane), rather than using much more modern, efficient technology. By replacing the
"obvious" steering wheels (visible to both pilots) with a set of tiny switches (visible only
to the one
who threw the switch), there's some communication being lost. Not all communication is personal,
nor verbal. Some is picked up from signals another person sends out to the environment, and is
received from the environment by the other person, almost unconsciously. This information is
crucial, and tends to be lost when we make the group workspace too efficient by eliminating
"crosstalk" and multiple channels of communication.
Balance cognitive overload with full accessibility of all information
There is a delicate balance here. For air traffic controllers, they have to deal
with lots and lots of
messages - cognitive overload. However, if they only get the messages that pertain to them (say,
they are dealing with plane X landing on strip Y right now, and only get info. from plane X), they
lose sight of the total picture. This can cause more mistakes.
Mistakes
Norman uses the example of a ship's crew. They communicate through handsets
and hear
each other talk. Crosstalk. Yes, they do make mistakes, and under pressure or confusion they
make more mistakes. But to eliminate those mistakes by eliminating the crosstalk means we
eliminate the learning process by new crew members, as well as the mentoring process by old
crew members who need to teach the newbies how to run the ship. "This shared communication
channel, with its shared teaching and correcting process, keeps everyone at a uniformly high level
of expertise.
Efficiency
Natural, smooth, efficient interaction should be the goal of all work situations.
However, natural
interaction is often invisible, unnoticed interaction. We don't know it is there till we remove it,
and
then it may be too late!
Disembodied intelligence
He's talking about simulations. Simulations are good for teaching. However,
they're difficult to
program because of all the physical constraints that are automatic in the real world, but have to be
factored in when designing a simulation. All too often, in the effort to come up with a programmable
solution, scientists simplify the situation, then solve a simpler problem, with the goal of solving
the
real, complex one later. I've seen how this fails (with atmospheric reflection, using lots of simulated
little mirrors - it simply doesn't work!)
Accuracy
It's not natural. Humans are organized around sunrise and sunset, stories, social
interactions.
Just how important is accuracy to our real daily lives anyhow, until we start introducing technology
that demands it? It becomes like Charlie Chaplin in Modern Times - going by the clock, with
studied movements, very dehumanizing.
Oral tradition
I disagree with Norman. He does not distinguish information transmission from
ritual
communication, as Pea does. What he says is true for information transmission - storytelling - but
not ritual communication. In the Sanskrit oral tradition, slokas are memorized in a very formal
fashion. You cannot substitute one syllable for another, because it will change the root of the word,
or it will affect the meter. This is true all through Tibet, Nepal, India - anywhere that's based on
the
Sanskrit oral tradition. Vedic (pre-Sanskrit) was not written - that oral tradition still holds, even
through they did come up with a written language around 1500 BC.
Chapter 7
There are some terrific examples of technologies that sounded smart but did not
work, because
they were not really suitable to the way real people organize things. Each helped, but had critical
drawbacks.
Wooten patent desk
Organizes piles of paper from on your desk to around your desk in pigeonholes
No place for labels; no way to organize them hierarchically
Limited storage - stuff that won't fit into, say, 100 slots is lost.
File cabinet
Hierarchical organization (like a gopher server)
Unlimited storage
Requires standard size paper
The deeper the folder in the hierarchy, the more time it takes to retrieve
Nonstandard paper size stuff is lost (check stubs, greeting cards)
Alphabetical schemes
All information is accounted for in one level (like a laundry list)
Unlimited storage (but dictionaries get very heavy!)
No cross-indexing (synonyms, thesaurus)
Need to be an accurate speller to use it
Language-sensitive (no "shch" from Russian, no "ch" from
Hebrew, etc.)
No linking of ideas
My research management product
I use file cabinets for papers, piles of paper on my desk with post-it notes,
piles of papers on
one shelf of my bookcase, for hard copy. The periodic backups of the hard drive are on a single Zip
disk. I use the desktop metaphor with folders to organize my files on my hard drive. For my online
stuff, I try not to be too "gopher server" or "laundry list" oriented, but I simply
cannot stay within the
2-click rule. Online works better than the Mac, because I can put hypertext links into my readings,
where I consciously relate (form an elaboration) between the new information and prior knowledge.
Norman doesn't mention hypertext at all.
Desktop metaphor
Mac interface is recognizable and useful because it's like a file cabinet
Easier to scan - click on folder to see what's in sub-folders
Same drawbacks as physical file cabinet - what folder contains what paper?
Ideas to organize the stuff by - file cabinets are better since you can <
Databases
Easy access - set up the search and it finds stuff nicely
Relational (search on related ideas)
Requires input to be standardized (like standardized paper for file cabinet)
Info. that won't fit standard fields is lost (long responses to open ended questions
in large
surveys)
Not intuitive - need to learn DB3 or Access - must talk like a computer
McGuckin's hardware store
Same idea as a file cabinet (hierarchical)
Add intelligent agents who know what's there and what to use it for
Needs a human being to make it work - can't be automatized
Cyberspace
Unlimited storage
Unlimited types of data files (gopher, ftp, www, text, graphics, etc.)
Not hierarchical - gopher server won't work - no spatial metaphor - navigation
is very difficult
Can't find things in 2 clicks - laundry list won't work - too big
Need sophisticated search engines and parallel processors to make the search
look seamless
If you don't know what you're looking for, it won't help you
Location depends on how we use stuff - use it a different way, and you'd expect
to find it
elsewhere (try to find Durkee's fried onion rings for stringbean casserole - is it a snack? a
vegetable? does it belong next to the fried shoestring potatoes? Guess again!) Nor does
organization remain fixed over time, especially as new classifications of knowledge arise. Lastly,
there's the navigational problem, which drives teachers crazy when they try to find lesson plans on
the 'Net. They give up!
Why don't these schemes work? Because human memory isn't hierarchical, following
simple
paths. It's "navigation by description" - describe what we care about, and it's there. You
need
multiple routes, by which the user can get to any piece of information, in a way most relevant to
him/her. This can't be replicated by normal mathematical logic, which is what programmers would
like it to be.
Basically, any time an innovation organizes something for you, it uses you. It
makes you think
mathematically, spell like the dictionary, organize stuff the way the inventor wanted you to, uses an
algorithm - not "your" way. So for every pro, there's a con. Plus, you lose information that
won't fit
into the innovation's structure.
Chapter 8
When a new technology is introduced (airplane, FAX, etc.), both the technology
and the society
have to adapt. People considered the telephone to be a broadcast medium - later on they realized
it was a means of personal communication. This mutual accommodation takes time. It also
involves a large change to the supporting infrastructure. Nuclear power may be cheap, but it still
needs delivery channels, and the building has got to be made as foolproof as possible - this takes
lots of time and money.
New technologies introduce new problems: invasion of privacy, issues of equity
between the
haves and the have-nots, sociopaths and hackers, and new forms of personal interaction. Do you
want someone to see you in the bathtub? Or do you want the videophone to look at a picture of you
in the bathtub while you are doing something else? The technology has the power to falsify, to
intensify your fantasies. Maybe it's OK for flight simulators, but just how much of a false image is
OK for you to portray, to put on your best image, to even project yourself as an avatar. Does the
technology have to degenerate to Dungeons and Dragons?
It's not a question of whether these technologies can be made to work. They will
work. And they
can even afford true reflective comparison (as well as mass deception) - what you will look like if
you lose that 20 pounds, change the color of your dress, your hair, etc.
Human beings can also be trained to work with the technologies - see what the
elementary
school kids are doing with computers today. Not like adults whose brains have gelled! We are
capable of learning all sorts of arbitrary means of communicating ideas, making representations,
and sharing them with others. We could also become like the Borg on Star Trek.
There are always limitations to the technologies, that don't interface well with
humans. Norman
talks about many possible ways the technology might change in the future; he should read
Negroponte's Being Digital and compare futuristic visions. Whereas Norman complains about the
limitations of the technology, like the graininess and jerkiness of virtual reality, Negroponte says
you should forget vr and putting yourself into some false environment - holographic images
projected around your room will fix the situation and everyone can watch them. But he also talks
about touch screens and how they are not sensitive to real human fingers. Whoever you believe,
Norman warns us to beware of the quick-fix!
And what about group interaction? We see lots of papers on CSCW. Yes, you can
draw stuff on
a jointly shared workspace. But what are the effects on the social processes of interaction among
the group? Norman says that when technology supports the individual, the individual can still
control the situation, and the technology/person pair can find some graceful means of interaction.
But when people work together, social tensions can arise, that can only be avoided by the good will
and cooperative attitude of the team. Add an inflexible technology, and you make this difficult.
Well, there are problems with e-mail, but I have seen lots of growth here at the
Ed School as
people learn to work with it, discover its affordances and effectivities. Maybe we are training
ourselves to think like machines and compensate for the lack of body language, but it certainly
empowers us to create a knowledge base and community that is distributed in place and time. I
am not such a pessimist!
Norman's last comments are that it's not just the technologies, but the people
who work with
them. Same argument as he used before - the entertainment people use the experiential mode for
entertainment at the cost of reflection; the programmers use reflection at the cost of experience.
How many of us can program the vcr?
Chapter 9
This chapter reminds me of Turkle & Papert's paper on epistemic pluralism
- the difference
between the hard and soft approach to technology. Basically, machines are good at some things(
good at calculating, bad at perception, people are good at other things (good at perception, bad at
calculating), and they are different.
Besides the hard-wired perceptual abilities, people also map problems back onto
their own
personal knowledge and experiences - mental maps, like Kintsch with the city maps. Humans use
stories; machines use logic. Logic abstracts the critical, quantitative aspects of a situation, and
provides general (decontextualized, mathematical) methods for reaching conclusions. Stories
emphasize the special aspects of the situation (contextualized, non- generalizable, situated), that
emphasize the human side of the matter. Since they emphasize the qualitative aspects, they are
not quantifiable, and will not have the reliability that a logical process will. It is like the difference
between qualitative and quantitative methods in research - I think you have got to have both!
Norman agrees.
Humans use language that has taken aeons to evolve to its current form - it allows
for ambiguity
and imprecision, and emphasizes ease of use rather than linguistical rigor. Languages vary
according to the shared experiences of the culture in which they are found. They are not easy to
analyze logically. They are "soft".
Technologies, too, are hard and inflexible, forcing humans to conform to their
needs, rather than
the other way around. The stamp machine, for example, doesn't have an "undo" command -
something that software developers have found that humans consider essential in good software
design. From the point of machines, people are vague, disorganized, distractible, emotional, and
illogical, whereas machines are precise, orderly, and logical. However, from the point of humans,
it's the other way around: people are creative, attentive to change, resourceful and flexible, whereas
machines are rigid, insensitive to change, unimaginative, and prone to "a foolish consistency -
the
hobgoblin of little minds". It's all in your point of view.
It's important to make the link with Turkle & Papert here, because their paper
deals with not only
the gender difference in approaching technology, but the whole arena of soft vs. approaches to
technology.
Chapter 10
In "technology is not neutral", Norman re-emphasizes the difference
between man/machine and
experiential/reflective. "Each technology poses a mind-set, a way of thinking about it and the
activities to which it is relevant, a mind-set that soon pervades those touched by it, often
unwittingly, often unwillingly" (p. 243). Amen!
Each medium has its affordances as ( Allen & Otto also say. Plus, Diane Ravitch
warns us
about the dangers as well as the benefits of the new technologies as they should or shouldn't be
used in the curriculum. Reading affords control of pace, mental concentration, and effort. TV affords
ease of watching, and a continuous flow of information that drives the senses experientially.
People are capable of reflection; machines are not. To reflect, a system must
have an internal
representation of knokwledge and the ability to examine, modify, and compare its representations -
he calls this a "compositional" representational medium. Paper and pencil can augment this
- TV
cannot. But reflection also requires the time and ability to elaborate upon and compare ideas - TV
cannot afford this. However, when used interactively, TV changes, since human beings can
transform any technology into a reflective one.
Technology can be used appropriately, as long as it conforms to humans rather
than the other
way around. This can be done, like hiking boots and pocket calculators. "The difficult problems
are
the social ones, not the technological ones" (p. 252) - but this is another matter altogether...
1. Thoughts
