Human Information Processing (Perception)

From CS 160 User Interfaces Sp10

Jump to: navigation, search

Contents

Readings

Pages 24-76.

Discussion Questions

  • How can the ideas presented in this chapter be applied to the design of user interfaces?

Mattkc7 - Mar 02, 2010 01:17:34 pm

Jason Wu - 2/26/2010 22:37:59

I'm a little bit skeptical about assignment actual numbers and equations to various processes within the human mind, but I feel that concepts behind the equations are very applicable to UI design. The authors state that very similar events occurring within the same perceptual processor cycle are combined into a single percept. If a designer wants to create an interface that provides the user with auditory or visual feedback, he or she must ensure that the tones, changes in color, changes in cursor, etc. either never occur within the same cycle or are different enough for the user to easily differentiate between. This is very important, since otherwise, the user could potentially miss a critical piece of feedback. For example, if a user performs an action that causes two separate errors in an interface, both of which issue the same warning tone at almost the same time, the user will receive one percept and will likely only realize one error has occurred.

Another concept that is very applicable to UI design is perceptual causality. The authors claim that the illusion of causality breaks down at 100 msec and even as early as 50 msec for some people. UI designers must take this causality breakdown time into account, since users of their programs will likely expect what they perceive to be an immediate response/output to their input, and the breakdown of causality will therefore be quite jarring for the users. To ensure causality, I believe that all interfaces should be designed for Fastman (i.e. the interface should respond within 50 msec), and hopefully with newer, faster hardware, computation times will no longer lead to a breakdown in causality for the user.


Richard Mar - 2/27/2010 12:23:56

The ideas presented in the chapter provide some useful guidelines for designing an interface. In particular, it discusses the limits of a person's memory and cognitive facilities with respect to incoming information and stimuli. By keeping information output in short chunks, the user can remember more of it faster and for a longer period of time.


Daniel Ritchie - 2/27/2010 21:22:12

I liked the level of abstraction at which this reading treats the human mind. By defining the "Human Information Processor," the authors utilize a mind/computer metaphor, but they don't take it too far. I've read cognitive psychology texts that have carried the metaphor into the "implementation details": for instance, equating working memory with a computer's primary DRAM due to the short-term nature of both memories unless refreshed (or rehearsed, in the case of the mind). I think this is excessive--after all, a computer isn't prevented from carrying out other tasks as its DRAM refreshes (whereas humans have trouble doing anything else while mentally rehearsing some information).

This reading also raised some concerns related to my group's project as it got me thinking about Fitts' law. Our app (and those of many other groups, I imagine) need to pack a lot of objects into very limited screen real estate. This will tend to make said objects very small targets for clicking/tapping and could slow down task completion times. I also wondered if occlusion due the "Fat Finger" problem is rolled into Fitts' law (as part of target size), or if this is yet another factor that could contribute to user slowdown.


Charlie Hsu - 2/27/2010 21:49:18

The reading's calculations of perception, cognitive, and motor time were useful supplements to thinking about user interface design. Keeping the illusion of causality is an important part of user event-driven interfaces, and the reading specifically mentioned thinking about perception times in determining how many frames per second a video should be shot in to maintain the illusion of motion. Likewise, we should think about perception and cognitive processes in our own UI design: if a process takes longer than the average perception time to complete, we should put in an activity indicator to notify the user of the computation. We should set iPhone widget animations to resemble times of perception and cognition; we don't want to create an exceedingly slow animation, but we also want the user to be able to recognize that his/her action caused an animation to fire.


Owen Lin - 2/28/2010 12:14:08

It is very interesting how the human brain and motor control can be analyzed and its performance be given in equations. By knowing these guidelines of human capability, one could design interfaces that minimize the fatigue in using the interface and concentrate on the content. For example, you could design your interface so that similar controls are grouped together to minimize travel time to associated buttons, and have animations in your interface run at 18+ fps to give them fluidity. You could minimize the interruptions from the interface (such as pop-up menus obstructing your content) because it drastically increases perception/reaction rates.


Matt Vaznaian - 2/28/2010 15:24:44

This was a very dense reading, essentially relating every part of brain processes and memory to mathematical equations. The ideas presented in this chapter relating numbers and formulas to human cognition. This is relevant to user interface design because when designing an interface it is important to keep in mind what kind of interactions the user is doing with the screen (or other input device). Interactions include typing, pressing buttons, moving sliders, etc. and where the interactions take place and in what order actually does matter. By applying the formulas presented in this chapter to an interface design one can make smarter implementation choices, allowing for better continuity and flow between interactions that are basically subconscious to the user.


Alexander Sydell - 2/28/2010 17:32:43

I found that using computer terminology to describe the inner workings of a human in the first half of the reading made it much harder to conceptualize. I feel that the authors should have used a different approach because, in the end, humans are not computers and describing them as such as unhelpful. However, the second half of the reading was much more useful for me. In particular, the various time measurements such as reaction time were useful to know in developing an interface. I believe the most useful section was on perceptual causality, as it is important to know that maximum time an interface has to perform computations and respond to the user in order for the user to perceive that he made the change. The under 100 millisecond mark is particularly useful on the iPhone as the device does not have a lot of processing power, and thus it's important to make sure the interface responds to user actions within that time.


Daniel Nguyen - 2/28/2010 17:34:17

Considering the human mind as analogous to a "Human Processor" is useful for user interface design because the design choices can be made to "optimize" the human processor, much as you would optimize a computer processor. This was already partially discussed in class in relation to Fitt's Law and the reasoning for the mouse being the best computer pointing device. Considering the principals given in the reading, it seems that unnecessary complexity in whatever tasks the user participates in leads to increased "cycle time" in the human processor. Thus, with regards to the Power Law of Practice, experimenting with a user interface that repeatedly presents relatively simple tasks that grow in complexity and build on previous tasks as the user interacts with the system more may present a way to make the usability of an interface uniform across a user's overall experience with it. This would essentially erase any sort of "learning curve" with the interface, instead making the system adapt to the user's experience instead of forcing the user to slowly learn the details of an interface.


Calvin Lin - 2/28/2010 18:28:31

One important aspect of creating a successful user interface is the ease of habitual learning and pushing tasks to long term memory. The chapter provides useful information that could help engineers make design decisions that would optimize the development of habits and long term memory retrieval. Questions to ask include: how easy is it to perform common ask? When a user wants to perform a task, how quick will a user recall the procedure, and how does our design help or impede this? It’s really important to be mindful of human memory capacity and the limitations of our senses. Requiring a user to remember an unreasonable amount of information in order to user an interface would not work well. Good designs provide visual cues that trigger our memory and remind us how to do a task without much effort.

The chapter is very thorough in describing several different areas of human perception and reaction, and our typical behavior and response time. This is very important when it comes to interface design because we are, of course, designing interfaces for people. For example, understanding how much information a user can handle at once and how much can be remembered is important. If the user is at the end of a 10 step process and then must recall what happened at the beginning but has trouble doing so, the designers should probably re-consider the interface and possibly dividing up the steps across pages or providing more visual feedback. At the same time, we don’t want to dumb down our interfaces too much where there are an unnecessary number of steps to perform a task. It’s crucial to get a strong understanding of users, especially since characteristics vary from person to person.


Jessica Cen - 2/28/2010 22:27:34

I agree with the authors on page 81 when they say that it is the failure of retrieving that causes someone to forget and not actual lost of the memory. I myself have trouble retrieving a specific memory from my long-term memory, and it comes back after thinking too hard. I believe that user interfaces must be designed in a way that the user doesn’t store how to do certain tasks in the program in his long-term memory. For example, when I am using MS Word, I don’t have to remember how to insert a clip art into the document because that task is easy to figure out. I just have to go to the Insert menu and find where “clip art” is. This action doesn’t require me to memorize where specifically in the program I have to click in order to achieve my task. Therefore, successful user interfaces must be easy to figure out and not depend on the user to remember them. To make them easy to figure out, user interfaces must be self- explanatory and simple, and designed in a way that it is easy to use for beginners.


Jungmin Yun - 2/28/2010 22:38:01

This reading is interesting but a little bit hard to understand because of lots of numbers and equations that I don't really understand the meanings of. I found this reading interesting due to the relationship between human brains and computer processors. However, since human brains are much more complicated than computer processors, I don't think that it's is a good idea to quantify everything and put them into some equations. This article says that visual memory has bigger capacity than auditory memory, but visual memory decays faster. For example, most programs have visual tutorials because they last longer than users' memory, and some programs use sound to notify users of errors or warnings because it lasts longer than just visuals without sound. Even though auditory memory lasts longer, I think that the auditory memory is hardly used in UI design. In my case, I usually try to see what's going on, but I do not really focus on listening because reading is more accurate and faster than listening. I think most people feel the same way I do.


Wei Wu - 2/28/2010 23:17:01

While the cycle time constants for perception, cognition, and motion that the article discusses can help to design interfaces that optimize user performance, i.e. the oscillator-based telegraph, the human-processor model also shows that the possibilities of what can be achieved by a computer--regardless of how great the interface design is--are still bounded by the limitations of the human brain. For example, the article discusses the maximum delay allowed to preserve the illusion of causality in simulating the movement of a second ball after some ball has struck it in pool. This delay is key from a UI design perspective for delivering feedback for some action in an appropriate time window to the user. However, if the feedback comes in the form of a new dialog menu where the user must choose something, even the fastest appearance of this feedback would not make user performance any faster due to the choice reaction time limitations of the user. He still would need time to assess the next step of his task, and it is this time that should really determine the maximum amount of delay the UI can afford to provide feedback to the user. Thus, taking all aspects of the human-processor model into account in UI design can help to prioritize the aspects of the design that need to be optimized, and single out the optimizations that may not be necessary.


Hugh Oh - 3/1/2010 3:16:01

Users have a limited rate at which they can perceive and interpret data. An interface only has to reach this breakpoint (e.g. videos only have the play at the frame rate at which we can perceive things). This also means that a interface has to limit itself in terms of rate of data being given to the user (e.g. scrolling text that an average user will be able to read).


David Zeng - 3/1/2010 3:21:30

Reading this paper was very similar to reading about caching for operating systems. In this analogy, the working memory is cache and the long term memory is disk. Just like how it is important that we minimize the things we need to keep on disk and try to optimize for performance on the OS, we want to minimize time spent on any UI, as well as try to minimize the learning curve or amount of things that need to go into long-term memory. By reducing the overall time needed to learn and use the UI, then users will have a much better experience and be more proficient. The reading mentions many small examples such as placement of keys or pattern matching that could be used to try and reduce the time it takes for the user to user the interface.

One thing that I would have liked to see is the phenomenon where time seems to slow down. It apparently happens during high intensity situations and the senses are heightened. I have heard that your eyes are able to see more frames per second, but I also wonder if the motor and cognitive functions are increased. It has actually happened to me once during a particularly important match for my school team in badminton. I remember seeing things travel slower than normal and even thought that it seemed slower than normal at the time. At the same time, I was able to move exactly as I wanted to.



Boaz Avital - 3/1/2010 3:51:56

So every user interface you design should go through three vetting processes: one for the perceptual system, one for the mtor system, one for the cognitive system. Another interesting approach would be to try and engineer a user interface directly from the list of principles that are important, going one by one and trying to maximize.

A point made in the reading was that to press a button you have to cock back your finger, push down and, and then pull back again. In iphones, there may be a good way to make pushing buttons easier. A swipe across a button in any direction, as opposed to a tap, could be a good way to streamline button pressing.


Vinson Chuong - 3/1/2010 10:16:27

The Human Information-Processor Model is an attempt to relate the familiar model of computation to how humans process information using empirical data. It's effectiveness as a working model seems to rely solely on how far we can push the computer metaphor. I believe that we should take the actual model with a grain of salt and pay attention to the empirical data.

With regards to the application my team is developing, most of our functionality relies on real-time input and output of multiple channels of information. One of my primary concerns is to design an interface that can present such information so that users can process and act on them.

Will our users actually be able to process multiple streams of information while performing a task? Are there ways for us to improve how the streams are presented?

Task Analysis and Lo-Fi Prototyping provide us with two sources from which to find answers. Empirical data on human cognitive performance gives us yet another.


Eric Fung - 3/1/2010 12:22:21

The section on time delay and its effect on perceiving causality was particularly interesting. It is extremely frustrating when you run into an error in an application and need to think back a few seconds as to possible actions that might have caused it, whether another open program might have had something to contribute, etc. The idea of constant and immediate feedback for the user has always been a major theme in the past readings. Now, the term 'immediate' has upper and lower time bounds. It seems that two events occurring within 0-112ms of each other can be considered to still be linked together causally; any longer and this illusion of causality disappears. So ideally, errors and alerts should be raised within this time window to let the user know what actually caused the error--more important than knowing there was an error to begin with.


Dan Lynch - 3/1/2010 12:43:12

Very intriguing article goes into the depths of the mind and cognition. An analagy between a computer and the the brain is made to effectively describe particular interactions with interfaces. The first portion of the writing dicusses memory, the motor system, and maps these to ideas concerning a cognitive processor. The article then goes into depth on mathematical laws that describe the limits of humans and their reaction times. Basically it all boils down to Fitt's law, which was actually derived in the paper. I do think that a lot of assumptions were made with the derivation, but nevertheless, it was interesting. I really enjoyed the positive note of our limitations: the Power law of practice, which utilizes the fact that perceptual-motor learning occurs---with repetition you will get much better with time and practice.



Arpad Kovacs - 3/1/2010 12:50:32

Although I found the quantitative modeling of the human brain via formulas/graphs interesting from a purely academic cognitive science standpoint, I think that this article would have been more useful for user interface design if the authors had given more explicitly perscriptive advice for how to leverage their findings. So I guess I'll try to distill the essence down to a few points:

  • Due to Fitt's Law, heavily used modifier keys (eg function key on calculator, or Shift key on keyboard) should be placed as close as possible to the nominal work area to minimize movement time.
  • Delays in human perception can be exploited to create illusions of motion (if rate is faster than cycle time of the Perceptual Processor), or causality (if the stimuli occur within 50 or 100 msec of each other). However, these delays also define the limit of auditory or visual input bandwidth, therefore sensory input should be
  • Reduce unnecessary stimuli/difficult decisions as much as possible to optimize user reaction time (don't add unnecessary distracting sound effects to interfaces, and avoid making user retrieve info from long-term memory).
  • If you want the user to remember something, force him/her to retrieve it immediately from short-term memory before it is forgotten. Also use meaningful and unique identifier names (to avoid interference in working memory due to the discrimination principle), which are easier to transfer and associate in long-term memory.

These observations help explain why some common paradigms in modern user interfaces work so well. Tooltips are awesome because a user can just hover over an unknown button and read a unique, meaningful message that does not interfere with other stuff in short-term memory, without the reaction-time penalty of having to retrieve this information from long-term memory. Dialog boxes which encode redundant information via color and shape (eg a green checkmark next to an OK button, and a red X next the Cancel) are also helpful because the cognitive system is parallel in the recognizing phase, but serial in the action phase (so don't offer to many choices either).


Tomomasa Terazaki - 3/1/2010 13:23:11

This was an interesting article. I think this article would be interesting even for non-computer science students because it talks about how human brain, or more specifically memory system, works and how it is affected in our daily lives (especially with technology and how computer scientists design their devises for whatever reason). I think it was interesting how it talked about all the different ways to get and store memories in our brain. The most common ways are watching, listening, and actually doing. Actually doing something is the easiest way to memorize something because the person is thinking and his/her body will memorize what to do. Also, if a person does enough of this it will be in the long-term memory, so soon the person does not even have to think when they are doing the action. For example, typing. In the beginning, most people use the learn how to type program to learn where each key is located and even afterwards have to look at the keyboard to type. However, after few experiences it becomes easy and will think it is crazy to relocate the keys in the alphabetical way because that will take too much time to type. However, it will end up being the same because even if the keyboard is changed to alphabetical order soon people will get used to it and type like they type now (even though there might be a small difference in time because the keys right now are located so it is easier to type most of the words). Also, what I liked about the article is how much of a time difference it makes if we change something we use in daily lives (the example and solution part). After reading that I thought about switching all the places where I put my plates and utensils. Right now I have my spoon on the other side form the bowls I use to eat cereal but if I relocate my spoons to be near my bowls, I will probably save a second or two.


Chris Wood - 3/1/2010 13:26:08

I found this week's reading very important to understanding how to best design a user interface for humans. There is usually a set of physical movements that best cater to the cognitive makeup of the human mind. A good success story is the navigation of the mouse pointer. Though the movement of the user's hand is perpendicular to the actual movement of the mouse pointer, the connection makes perfect sense to our minds. We should look to working examples such as this when trying to design new physical navigation techniques through an interface.


Joe Cadena - 3/1/2010 13:57:18

From the reading, I learned that the 100 msec cycle time of the recognize-act cycle from the cognitive processor can be shortened through practice. Therefore, I believe building an efficient user interface would include a design model similar to what the target user is used to interacting with. More user recognition and less chunk learning would enable the user to quickly grasp the functionality of a new user interface.


Michael Cao - 3/1/2010 14:29:42

After reading about the model human processor, there seems to be a lot to consider when designing a user interface. Human capabilities can often be very limited so those limitations have to be taken into account when creating UI. Some of these limitations can be memory, distance, speed, and ability to multitask. In the end, these are all ideas involving the ergonomics of designing a user interface.


Geoffrey Wing - 3/1/2010 14:29:44

This week's reading was an interesting read - the analogy of the human cognitive system to a computer was not something I had thought about before. It was especially interesting to me to see quantitative analysis of different functions, such as the working memory capacity and the cognitive processor cycle time.

With a good understanding of the human information processor, you can design better user interfaces. With Fitt's law and human reading speed, you can time prompts to allow users to fully understand what is going on and make an informed decision. Memory capacity is good to know, if you want to have users remember information. However, I don't think users should have to remember too much information in an application, and for screen prompts, I think it's best to let the user decide when they have finished reading. Timing prompts to automatically advance can get tricky, since users may have to step away from the computers for any number of reasons.


Aneesh Goel - 3/1/2010 14:34:02

The reading provides two valuable sets of ideas to consider when designing a user interface. The first is the qualitative assessment of the model of the mind as a processor; when designing software, thinking of the human interface in terms of an API being used by a different kind of processor makes it easy to shift gears between front- and back-end work; providing rough equivalencies between psychological capabilities of the mind and hardware capabilities makes it easier to keep, for example, limits of working memory in mind and design around it, such as providing more efficient chunking and fewer chunks the user must focus on. By using this model as a reference sheet, designing software for users becomes much easier.

The second is the set of equations allowing quantitative analysis of performance; one can predict efficiency gains from various designs before implementation and testing, streamlining development significantly (lo-fi prototypes are great for many things, but the 'human computer' introduces enough lag to make anything involving timing impossible to measure without higher fidelity examples). Fitts's law we covered earlier, but the power law for learning provides a curve to model user experience without needing to actually have testers become familiar enough with the system to actually be experts; reaction time measurements and speeds for the three stages of human processing make determining the timing for events less guesswork and more science. All together, the paper brings together a lot of useful metrics for analyzing design ideas before even reaching the prototype stage.


Darren Kwong - 3/1/2010 14:43:31

With all of the principles and laws governing the human information-processor model, it is possible to quantitatively compare and analyze user interface designs in terms of human information-processing time. This can improve data flow, the placement of objects such as buttons, and the implementation of task-oriented features. However, it can be initially be time consuming to properly do such analyses. How much improvement can be expected from using these ideas in comparison with more arbitrary designs based on iterations of (lo-fi) prototypes with user interviews?


Alexis He - 3/1/2010 14:59:13

On Physical Matches on pg 66, the author states that "If the user has to compare the stimulus to some code contained in memory, the processing will take more steps."

We can learn from this rule of thumb when designing systems to create analogies with recent images or processes. For example, I noticed that a lot of the commercials during the Winter Olympics had a winter olympic theme to them (snowboarder advertising a product) so your brain doesn't have to context switch out of winter sports to make the connections. Similarly the winter olympics video feed page can utilize this and make the icons recognizable by winter sports fans.


Victoria Chiu - 3/1/2010 15:08:48

Fewer options take fewer time for a user to choose. Naming is probably very important to interface design; in the example of "light" as opposed to "dark", if the user who is retrieving the file name thinking of "light" as opposed to "heavy" instead of "dark", the user will not be able to recognize the file. The idea is that if a certain thing has a stronger association with other thing than the retrieval cue, the information might not be functional anymore even though the information is physically present. Choosing appropriate names, which is the retrieval cues, is important.


Saba Khalilnaji - 3/1/2010 15:12:44

It seems the weakest link in data processing eye movement which can take 200ms. Perhaps any type of user interface that requires a lot of eye movement, such as with very large monitors or a UI where data or input is spread out will sufficiently decrease the efficiency of the system. Furthermore it is quite interesting that auditory memory lasts about 8 times longer than visual memory but stores only about a third of the amount of "data". An interface that has auditory output can expect an accurate response about 2 seconds after it is put out, this might be useful for applications such as video games. If a character is forced to listen to a long dialogue for whatever reason how much of that information are they actually getting? Visual output might be preferred for longer data with audio for long ranged data. Lastly I found the equation for the reaction time to signals of unequal probability to be really interesting. This can be used in so many ways to improve so many UIs!


Brian Chin - 3/1/2010 15:33:58

The reading provided an interesting way to think about people as computers, complete with processors and memory. In terms of using this information for making user interfaces, I think it provides broad guidelines on what humans can be expected to do. In the reading it describes how fast humans can react to stimuli, how fast they can process information, and how fast they can do a response to a stimulus. This can be used in UI design in that the design should not expect or require users to move faster than what is physically capable. An example is that if a designer was creating a game, he shouldn't expect a user to have 10ms response times. The information in the readings can be used in other ways as well. The reading talks about how to calculate the speed of how fast humans can do certain actions, such as pressing buttons. This information could be used to create things such as button layout that could be optimized to be as fast as possible, depending on the task.


Mohsen Rezaei - 3/1/2010 15:42:50

Taking into account the probabilities and action/reaction calculations for movements, decisions, and/or thoughts we would be able to make interface designs that would give a user just enough convenience to make them enjoy and accept our "well" designed UI. In a "well" designed UI we could help users decrease time spent to make decisions by, for example, color coding texts or controls existed in a view. Furthermore, we could help them make right decisions by giving longer time-out on alert notices. In addition, some people have better output and faster performance in finishing a task when for example buttons involved in that task are well positioned in the view so it will actually help the user make decision about what need to be pressed after the one before it. This is an example of helping a user decreasing "Motor Skill" time. Another example for motor skill access time is the well-known delete button in the editable TableViews in iPhone. As a n improvement the red Delete button could be appeared close to the red minus sign button. Access times, whether it is working/long-time memory access time or movement time, are analogous to the I/O of a computer. It would be faster to read some data from the Cache or Memory that straight from the input devices like Hard Drive Disks.


bobbylee - 3/1/2010 15:50:07

In the reading, Variable Perceptual Process Rate Principle mentions that the perceptual processor cycle time tp varies inversely with stimulus intensity. For instance, when we have an alert button, we could make the button a high-contrast stimuli, as it will require less Porccessing cycle time for users to recognize that it is an alert button. In addition, given that the half-life of the Visual Image Store is a lot shorter than that of the Auditory Image Store from the reading, when we design an interface, which we want the user-interface to be more memorable, we could add some sounds to the user-interface. When you play the video games, there is always a sound effect responding to certain actions done in the game. For instance, when you kill some monsters, the game might play a sound of victory. Since the sound will stay longer than an image in your mind, it will make your experience more enjoyable as your happiness might last longer and have a larger intensity.


Divya Banesh - 3/1/2010 15:50:54

I was interested to learn about the different rates of decay for visual verses audio memory span. It seems that visual input decays a lot faster than audio input and this is probably why people can serf the web quickly or flip through several windows quickly and understand what they're doing. However, if audio stays in the memory so much longer, it might be a better tool for learning procedures. This might be something user interface designers can use so that beginners can learn the setup of a product better with audio tools.


Spencer Fang - 3/1/2010 16:08:30

The problem space principle can be applied to how users interact with GUIs. The principle states that the way people solve problems can be described by the states of knowledge, operators for changing states, constraints on applying operators, and control knowledge for deciding which operator to apply next.

A user might use our BART app to figure out how much it costs to go from Downtown Berkeley to the San Francisco Airport. He might look at the app's title screen and see all the available buttons, and know that at this particular state, the operator is (TAP button). The relevant control knowledge in this context is that tapping on a button has the effect of moving to a menu associated with the button's text label. The user will tap on "Calculate Fare", because this is the next desired state.

Once at the fare calculator menu, the user will see a UIPickerView with compartments for origin and destination, and a button for Calculate. The operators here are (ROTATE picker), (CHECK picker), and (TAP button). The user will alternate between ROTATING and CHECKING the picker until the desired locations are selected. This demonstrates the user's control knowledge as well as constraints on operators, since he knows that (TAP button) on Calculate will not have the desired effect unless the correct origin and destination are inputted.


Sally Ahn - 3/1/2010 16:33:10

I am impressed by how the Model Human Processor abstracts the complexity of the human mind, but nevertheless captures all the essential details required for designing user interfaces. By organizing its function in terms of memories and processors, it allows us to consider the "cycle time" of the user's perception and response. The reading provides a good example of how it can be analyzed to explain a user's ability to respond to certain stimuli (for example, identifying the correct number of clicks when presented at 10 clicks/sec vs. 15 to 30 clicks/sec). Obviously, such information is important for user interface designers who must ensure that the system does not exceed these thresholds. It is interesting to note that this parameter is not constant, but varies according to the intensity of the stimuli. This raises the question of how such "intensity" of stimuli is measured.


Yu Li - 3/1/2010 16:37:42

The human mind can be seen as a computer system consisting of a set of memories and a set of principles (principles of operation). It also consists of three subsystems: perceptual, motor, and cognitive system. The perceptual system is made up of sensors and associated buffer memories. The motor system is makes up responses, and the cognitive system receives information from the visual senses.


Raymond Lee - 3/1/2010 16:39:29

Taking all that is discussed in this reading into account, there's some UI improvements that we can make. We can make sure that the user doesn't have to remember too much information on his own. We can also make sure that the user isn't presented with too much information too rapidly, whether that be visual or auditory information. With Fitt's law, we know to position widgets in a way such that the user won't "miss" hitting the correct widgets with decent speed.


Conor McLaughlin - 3/1/2010 16:44:05

The article itself seemed a little overly dry to me, but I really enjoyed the section on the Power Law of Practice and the breakdown of which keys are reached by which fingers and how often those keys are used in combination. The idea that an alphabetical keyboard would only be 8% slower than the current arrangement of Sholes seems shocking to me. I remember from lecture that keyboards were made with the current arrangement to make it an easier transition over from typewriters, but if more exhaustive studies had been done on what the ideal arrangement would have been, perhaps typing could be much faster than it currently is. I didn't find much really to argue with in the article, but I definitely was taken back to that balance of innovation versus convention, which would be more marketable with a less steep learning curve in a product. However, the idea of being able to type even faster than I am currently capable of is a tantalizing one.


Jeffrey Bair - 3/1/2010 16:51:46

The idea of associating memory with certain things in long term memory in order to better remember the short term ideas can be used to help design user interfaces. Since items in long term memory are often things that have been repeated often, we can use this to our advantage when designing. For example, google maps is a sit that is used by millions and many people are already familiar with how it works. If we design a sort of locator for our user interface, having something similar or just importing the google maps directly and adding our own locations on top of it is something that will allow users to be familiar with the design and link it to their past experiences with google maps.

Along with this design choice, we also need to keep in mind that the functions of these items that are associated with long term memory also work in a similar way. If they look similar but are completely different in terms of how they are used, it can often lead to confusion. This is the concept of interference in long term memory where associated items act differently and can corrupt a users memory of how things work. For example, pasting in emacs is Cntrl+y whereas normally it is Cntrl+v even though they typically serve the same function. This can create problems where you try to Cntrl+v in emacs and Cntrl+y in other programs. Having a similar setup to previous iterations is a very important aspect of designing user interfaces for people already used to a certain way programs work.


Linsey Hansen - 3/1/2010 17:00:14

So I really didn't like reading this at all for the first 25 pages or so, mostly because it is old, and while I was already really family with most of the material they were giving from psych and cog sci courses, the way they presented the information was just gross for me to read at times (or at least hard to visualize since I already have a different way of thinking about it). Anyways, after thinking about this and being annoyed at the authors for giving me all of these numbers for movement times, listening/reading rates, etc, I realized that I was doing that thing where people generally start taking stuff for granted after it has been around for a while and thus dismiss it as trivial common knowledge whenever they hear about it. So after this realization I started thinking about how much work must have gone into figuring out what movements would be easiest for people motor-skills-wise when creating some of the original interfaces. While we did sort of cover some of this in lecture last week, I never really considered how important it is to research things like how long it takes to make an accurate movement, since doing this sort of research could have helped windows with their bottom left pixel problem.


Angela Juang - 3/1/2010 17:00:30

This paper analyzes people's abilities and natural tendencies in a way similar to looking at how a machine works. However, while it might be a pretty good model, I'd be interested in seeing how it deals with conditions like disabilities, personal likes/dislikes, talents, etc. People tend to react better to things they personally enjoy or are interested in (for instance, they remember them better, and are more inclined to learn about them), and worse to things they don't enjoy. Similarly, certain disabilities or talents can make people's reactions to certain interfaces unpredictable. I think in designing an interface, it's important to take those factors into account because you should be trying to accommodate as many users as possible in your target segment.


Richard Heng - 3/1/2010 17:06:08

The effectiveness of chunking, combined with the proposed infinite capacity of long term memory could be useful for large, complicated systems. We can partition systems that display large amounts of information to have distinct domains, and have the information be categorized into those domains. This can be especially effective for interfaces that undergo repeated uses, because the working memory of each use can eventually be encoded into long term memory. This is of course different from procedural memory, which does not need to be in the consciousness.


Kyle Conroy - 3/1/2010 17:09:00

When creating a user interface for some application, a designer must remember the abilities and properties of a human mind and body. However, the designer must also eschew the tendency to over analyze a design, clouding the main idea of the interface behind piles of small details. Moran and Newell go into great detail discussing conceptual models for the human mind and body, crafting equations, laws and theories. While this level of discourse is appropriate for an academic exercise, I don't feel that it's necessary when creating a high level user interface. I think a designer should have these ideas in the back of his head, but focus instead on bridging the gulf of execution.


Andrew Finch - 3/1/2010 17:14:56

While this reading contains a great deal of interesting information regarding the abilities and limitations of the human brain to process data and react, it is difficult for me to see it being used in the design of most simple user interfaces. The information presented in this document is extremely dense and technical, and a great deal of time and energy must me put into interpreting it, and reducing it into usable guidelines for designing an interface. It seems to me that in most cases, it would be much easier to simply brainstorm, produce lo-fidelity mockups, test them on user gorups, and repeat this process over and over again until the design has been refined and satisfies most users to an extent that is acceptable. The only place I see for the type of data that's presented here is in the development of interfaces where ease of use, and speed are extremely crucial, such as medical, military, or aviation interfaces.


Nathaniel Baldwin - 3/1/2010 17:19:19

This reading struck me as a bit off relative to what I've read in contemporary psychology textbooks elsewhere in my Cal career - but it may just be vocabulary, and it certainly may be that things have changed in the last quarter-century. Anyhow, at face value: If this article had useful ideas about HCI, from my perspective they were hidden behind an obsession with the quantifiable - reminded me of Google's "41 shades of blue" dilemma. One thing, though, that sounded sensible was towards the end of our assigned pages, when they talked about having multiple, equally-probable actions occur (and require input), vs. having most of the inputs require just a couple of the many available actions. To me, this implied that one should strive to group the most common (related) inputs together, and to try to minimize the variance in necessary input. This reminds me of how Photoshop groups its toolbar - a limited number of options, all together, with more specific, less common options as sub-options. Makes sense.


Andrey Lukatsky - 3/1/2010 17:23:20

Taking into account various traits of the human mind is probably one of the most interesting aspects of this course to me. From the reading it seems like we have a fairly decent set of well-defined properties of the mind. I'm wondering whether computers can represent humans navigating an interface. Something like this seems like it's a good way to automate UI testing. Perhaps we can discuss in class the various research going on in this area.


Wilson Chau - 3/1/2010 17:41:42

Professor Hartmann talked about the topics of this reading a little bit in class and that is the time it takes for humans to detect things like delay and also things like Fitt's law. This reading really went into depth about how humans perform in interface activities. It talked about how humans get a lot better at activities at first and then reach a plateau and really gave a lot of information that would be very important to someone designing an interface. It is important to take into account the things in the reading because it allows the designer to play to the strength of humans and stay away from some of the weaknesses of the human mind.


Mikhail Shashkov - 3/1/2010 17:44:23

I thought UI analysis was getting a bit mathy when I saw Fitts Law..but now...sheesh.

Anyway, these are all very interesting facts and calculations that seem like they would be usable on a need-to-know basis; clearly a good UI can be designed without knowing any of these numbers.

For one example of one fact that SHOULD be utilized, the chunking and magical 7 seem like a good principle one could employ when trying to orient a new user (having a tutorial with more than 7 chunkable things to employ is a recipe for disaster, especially if the tutorial is not easily reaccessable.


Brandon Liu - 3/1/2010 17:45:35

The most interesting part of the discussion was how a sequence such as "BCSBMICRA" was much harder to remember than "CBSIBMRCA". Given the year this article was published (1983) all of CBS, IBM and RCA were significant "chunks" in long term memory. Personally, I found "BCSBMICRA" just as easy or easier to remember, it could be split into the chunks BCSB and MICRA. The fact that memories and information can be encoded this way means designers can "get something for free" when making user interfaces. A naive example is the hotkeys in photoshop: Move is 'V', while Lasso is 'L'. Rather than have to remember a character for the twenty plus items on the toolbar, a user simply remembers what it is they want to do, and then they need to hit the character that appears in that verb. The acquisition time for these hotkeys is much less than is typical for chorded hotkeys such as those involving Ctrl, Shift, Alt and numbers.

Related to the discussion of long-term memory, a really interesting paper on the detail of long-term memory is at http://cvcl.mit.edu/MM/. In this study, participants viewed 2500 images, and then were asked if they had seen a certain image as opposed to another one. The item was juxtaposed with either an object from a different category, the same category, or the same item in a different state. In all the conditions, and most surprisingly the last condition, participants performed at around 90% accuracy (baseline is 50%). The article gives good evidence that some intuitive assumptions about the resolution of long-term memory may not be completely true.


Long Chen - 3/1/2010 17:48:31

How can the ideas presented in this chapter be applied to the design of user interfaces?

A detailed understanding of how the brain functions (relative to a computer) will help designers streamline the user interfaces and optimize for brain processes. The reading for today was extremely dense and I had trouble understanding most of the parts. The general understanding and key takeaway I gained was that simpler = faster. The brain only has the capacity to process so much information, for example the eyes with the forea can only focus on a few degrees of vision.

User interfaces must be designed with as little clutter as possible. The brain should only need to focus on one central item at a time and that object should be the most vital to the program or the task at hand. Other options and functionalities should be hidden away, available if desired by the user. The applications that try to do too much all at once are the ones that have more negative reviews because they overwhelm the users due to the reasons discussed in the reading - any one of the human senses can only process a few items at once (although at very fast speeds). By focusing on that idea from the beginning of the design process, an efficient user interface can be made to be both soothing and productive.


Peter So - 3/1/2010 17:50:22

Modeling human performance as a set of three processors is a fascinating idea. Being able to quantify the cycle times of each type of "processor" cognitive, perceptive and motor allows the designer to take advantage of the illusion of causality to shape the user's experience with the product. The example cited in the text mentioned simulating motion by displaying still images at a rate above the perception cycle time of 100msec so that the user would perceive the multiple pictures as one continuous event but now imagine using the same principle to simulate cause and effect relationships to teach the user about real world cause and effect relationships. Time can be used to associate events with one another or even dissociate them. In context to designing user interfaces the human perception cycle time must be considered when you want to show the interdependence between multiple events.

In the same thought applied to motor cycle time, a neat idea would be to have a configurable interface with movable buttons so that the user can group buttons frequently used objects. It would be interesting to see if this was done in an open source context and see if the layouts converge to a single optimum orientation. The article is thorough in describing performing tasks objectively but doesn't explore the variation introduced by personality and prior user history.


Esther Cho - 3/1/2010 17:50:36

The chapter gives minute details on the limitation of the human sense, such as visual or auditory response. When designing an interface, designers can take these limitations into account (like we've seen in lecture with Fitt's law). It reminds me of many flash games where the difficulty involves speeding the gameplay up. At what point does the difficulty become humanly impossible? Some of the issues the chapter brings up such as the fusion of clicks seems like it would not apply to design (who would design something that involves listening to a sound within 100msec?) but designers should consider other limitations (like response time, what kinds of content require less access time).


Weizhi Li - 3/1/2010 17:51:02

The author suggests that the slow and sequential cognitive processor requires the designer to consider parallel operation to avoid overload the user’s cognitive system. The article mentioned the word "sequential", which is vital that the brain shouldn’t have the attention interfere among several tasks. For example, you can’t talk to others while listening to the music because the cognitive system is overloaded. Although it is like a "parallel processor," humans have short attention spans when using certain applications and want instant results. So applications should provide different outputs that can give user some kind of interactive feeling and avoid external decision-making interrupt to slow down the user’s process.


Jordan Klink - 3/1/2010 17:55:02

I found the reading to be a very interesting attempt to quantify the human brain. Whether or not it succeeds in producing a successful (or one relevant to my interests as a user-interface designer) analysis, is of course up for debate. The primary flaw from the reading is a result of the the fact that it analyzes the human brain with the "model" human in mind. This is an extremely general method of analysis, and in fact has very little relevance to my interests. This is because whenever I design an application, I am always focusing on a specific target user, not a "model" user in the general sense. I almost never want to develop an application for all users in mind, only a very specific user audience. Hence, for my target user base, the statistics and information relating to human performance found in the reading's research will more than likely be skewed in comparison to how users only in my target focus will perform. This is not to say that the results found in the reading are useless, only that they should be taken with great skepticism since it is overly generalized.


Richard Lan - 3/1/2010 17:59:42

The authors of this article provide the treatment of the human mind as an information processor, and use this model in order to analyze human computer interactions. One point they tried to convey was that humans are fundamentally limited in the tasks that they are able to accomplish. These limitations mainly come about due to physical restrains, such as the amount of time it takes to move your hand to a target, and the frequency at which your eyes can process a series of changing images. In a sense then, user interface design is merely the art of figuring out how to create the illusion of direct manipulation by exploiting these human limitations. The human senses are gamed. For instance, if the human eye can detect up to 10 distinct screen configurations per second, then software designers creating an animation for screen transitions know that they must make the transition at least this rapid.

Another aspect of the article was the ease with which humans can accomplish certain tasks, based on the amount of practice they have received, and the ease of the task. Earlier, we were told that the fact that a user becomes accustomed to a system through practice does not mean that the system has a well-designed user interface. However, the fact that the amount of time a task takes to complete decreases with practice may be used to design systems that must be used very often, perhaps on an everyday basis. Because the user would be getting so much practice using it, the interface designer could reasonably sacrifice some of the system’s usability for other aspects of the design. Additionally, because one of the main systems in the human processor model is the motor system, designers must take into account whether a task is physically possible. Take the iPhone as an example. It has a relatively small screen compared to most other user interfaces. This limits the type of apps that an iPhone user would find useful, due to the fact that there is limited real-estate. Therefore, the designer must account for how precisely the user can perform tasks using a touch interface, which puts a cap on how much content a screen can hold, and how small each individual object may be.


Jonathan Hirschberg - 3/1/2010 18:04:35

Sensory cues that were present while it was being encoded can retrieve the information from long term memory. This is what happens when words or phrases bring back traumatic memories. Someone may be fine before, but a word just brings up a traumatic memory and they feel bad again. It brings back the emotion with the memory. Similarly, if you want users to recall facts, encode them with sensory memories and then the interface should use those cues to bring back the information. There should only be one way of doing things as another article said, so that you can practice it repeatedly and eventually do it quickly and instinctively. One thing that interferes with practice is to make a new editor that's different from the old editor. To take advantage of practice, make the new editor usable in the same ways as the old editor. That way there wouldn’t be this interference effect and the user can learn the new editor’s functionality without having to relearn its interface first.


Jeffrey Doker - 3/1/2010 18:07:08

Last week I said I wanted to see more quantitative analysis of human-computer interaction, and this reading really delivered. I am glad to know that serious scientific study has been devoted to measuring all the minute aspects of HCI. I am still curious, however, about how far these metrics have probed. This paper discusses mainly surface level cognition (physical response to stimuli, memory), but I wonder how these metrics (or other metrics) might respond to analysis of mathematical learning, intuition, and innovation. User interfaces typically include physical computer interfaces, however I argue that a mind interfacing with an abstract mathematical construct is just a different sort of user interface.



[add comment]
Personal tools