Sunday, September 23, 2012

Distributing Cognition with user interface

User interfaces can help people distribute their cognitive load.  Interfaces can
encourage participation, scaffold learning, show differences that matter, or only differences that matter, convert slow calculations to fast perceptions of the idea, support chunking, increase efficiency, and facilitate collaboration. 
Tetris, study done by David Kirsh, UCSD, research on players.  Moving and rotating pieces on the screen may seem like a waste of time--people actually move the block more than they actually need to, but as you become an expert, Kirsh found that experts relied more heavily on the cognitive effort of moving things in the world--became a bigger task. 

From the learning sciences, the Montessori blocks give representations for numbers as physical estanciations.  Can the transparent representations scaffold learning, by making things concrete. 
These are good examples of the power of representation.  Good representations show all the relevant information, with nothing extra.  They enable comparison, exploration, and problem-solving.  They should enable the kinds of tasks users would like to do. 
London underground subway map--first map to abstract the layout of the map from the physical geography.  The Underground map designers realized that the common task of riders is to get from a to b.  The physical details were more information than they needed and made the task more difficult.  The detail was stripped down, making it easier to get between connections.  They also represented the distances far away as less relevant, the suburban areas were squished into the map.  For someone aming decisions of where to get off, or the compromise with the distance between stations.  You might believe that  Nearly all representation design is about fitness to task. According to the London Underground map, a good representation is tight to the task the user wants to perform.

Temperature map from the WEather Underground.  The temp is drawn right on the place where the temp exists.  What are good or bad problems?  If you want to know what the temp is along the coast, and don't care about town, this gives good information.  Or, it gives good information about general temperature.  Every temperature is shown identity, so you need to scan.  If the temp color or size would correspnd to the weather,
Edward Tufte books on color and design.
The depoth below sea level is represented by ROYGBIV.  this is not an ordering of colors, it is substitutive, we don't automatically have a more or less than value for color, so the individual chunks of a depth pop out.  The yellow pops out.  If you are trying to get the sense of the depth of the sea, what would be better to use color as a representational view.


His redesign:
Everything above sea level is brown, similar to real world.  the water is blue, and the intensity of the color changes with depth, the deeper blues are darker blues, which corresponds to our intuitions.  The idea that water by the shore is pale, this idea leverages that knowledge we all hold.

What makes a good representation is tied up with the task of the user and what the user's expertise is.

Chess- exemplar domain to understand== 1971 Chase and Simon
hypothesis-experts have memory for board, or have their ten thousand hours
experts are better at remembering the configuration of the board  ONLY if it is an actual game.  If you arrange the pieces on the board in a way you don't play chess.  This relates to the ability of experts to leverage their knowledge of the domain.
Chunking interfaces
ideally should make it easier -- interface designers can we make interfaces more chunkable and place a lower load on our memory.  Bill Buxton looked at moving test on locations on a document--keyboard shortcut to cut and paste--that is 3 different operations, and if interrupted, you might forget what is in the clipboard.  The pictures is worth 10,000 words.  should we represent information visually or textually.
It depends.  One time visually is more effective is when you can convert slow reasoning tasks into fast perception tasks by making them visually salient--London underground map and the good coloring map.  Tables of numbers-can be difficult to see trends, but visually including outliers becomes automatically visually salient.

Worth 10,000 words--interchange ability? how can we establish the informational equivalence.
informational equiv is not necessarily equal to computational equivalence. 

example
A pari of parallel lines is cut by a transversal. 
sentence and the picture are equivalent, but the pic requires less cognition

Distributing cognition for group work Ed Hussion on airplane cockpits
the artifacts in the cockpit help.  on landing and takeoff, the speed needs to be accruate-most important
plastic indicators called bugs around the indicators--to mark what events will cause a change in the wings, etc.  By making the triggers for what will make the action manifest to everybody, these artifacts can help increase collaboration and safety.

With DC interfaces can also increase efficiency using diagrams, and facilitate collaboration, as in the cockpits.

Poor representations are unfortunately common but can be deadly.  When you enter a password and are told that it is not valid.  It should tell you the requirements for the password the first time.  A message like you must fill out all fields is ineffective.  It might show only the parts you missed, or show the part you need to fill out.  If the information that is needed is not present, users get confused.  Ebay funding confirmation page, you will see if you switch to a credit card.  One option is pay with bank account and one is pay with credit card.  The bright one is pay with a bank account to encourage users to use this.  The other one looks like it is greyed out, but keeps users from having ebay pay their credit card fee.
In outlook, gives a duplicate contact detected,  To streamline the information, it can show the final information and what the changes will be.  Should show both old and new information in same format. 

Further reading
Don Norman, Things that make us smart
Ed Hutchins, Cognition in the Wild
Herb Simon, Sciences of the Artificial.


No comments:

Post a Comment