Experience #1: Ford Sync

SI 688 Term Paper (Fall 2010)

Student ID: 55634294

Experience #1: Ford Sync

Scenario

Sync is voice activated system available in some Ford Vehicles that allows users to link phones via Bluetooth phone calls and music playback. It also has a navigation and information system called Sync Services. Sync Services failed to work at all on one particular occasion. I was near Grand Rapids looking for a particular location and connected to Sync Services in the car. Normally, the voice recognition works fairly well, but in this instance it never recognized the actual city I was looking for (it was just outside the city). So deferred to looking near Grand Rapids. Even then Sync informed me that "We found several hotels near Grand Rapids" — and proceeded to name two. All through the process it was impossible to tell how to cancel, go back, or how to navigate through the various options at hand. Eventually I just gave up and used Google Maps on the phone directly (after pulling over, of course…)

Concepts for Analysis

Concept 1: Mental Models

Mental models are an important aspect of using any device or system, as they help "interpret and organize new information" (Norman, 1998, p.163). A schema is a collection of associations that helps a user use a system. Often metaphor is used to related a presumably more familiar schema with an unfamiliar one to make learning a new system easier, i.e. the term "BioLibrary" used to describe a DNA data repository hopes that the schemas associated with a traditional library will help patients know generally how the repository works. If no mental model is provided, either through some kind of schematic drawing or diagram, then users tend to make up their own, forming "theories (mental models) to explain what they have observed" (Norman, 1998, p.39).

Concept 2: Knowledge Transfer

"No learning occurs in the absence of prior knowledge" (Norman, 1998, p.153). Whether or not this knowledge is useful for learning new tasks is not always clear. Sometimes prior knowledge causes transfer, where as suggested the old knowledge carries along and assists in learning the new material. There is also interference —either proactive, where previous knowledge gets in the way or learning something new, or retroactive when new knowledge blocks recall of previous knowledge, respectively.

Concept 3: Working memory

Working memory is the brain's information store that acts as an interim between perception and long term memory. One model of working memory divides its function into a "visuospatial sketch pad" (Wickens, 2004, p. 129) for visual imagery and a "phonological loop" for verbal information. Our brains have limits when it comes to what persists in working memory in terms of capacity, duration, and similarity of the items being stored. The limits of working memory are "7±2 chunks of information" [ibid], and these items fade from working memory at an exponential rate (Norman, 1998, p.157) if not encoded for long term storage or rehearsed and refreshed in working memory.

Analysis

Two of the main functions of Sync are effective because of figural abstraction transfer (Norman, 1998 p.167) — prior abstraction knowledge allows the use and learning of a new system. Sync is effective for making phone calls because it relies on the existing schema of a phone. Even though the control is by voice, most people already know the words associated with making phone calls: "Call" or "Dial". This prior knowledge has a positive influence in that it transfers and facilitates use of this new unfamiliar system. The same applies to using Sync as a music player. Typical commands for anything from a tape player to iTunes are things like "Play Music by Artist Dave Brubeck", "Pause", or "Next Track", and you can use these commands with Sync.

Sync tests the limits of human working memory is when it's trying to help. If you fail to issue a command it responds with "You can say USB, Line in, Bluetooth audio, phone, read messages, vehicle health report, or help. If a route is active, you can say route status, route summary, cancel route or voice guidance on or off". To someone unfamiliar with these commands this list can be hard to follow: 1) because it exceeds the "7 ± 2 chunks of information" (Wickens, 2004, p.129) — depending of course on how well the user can encode what they are hearing. 2) Because "decay and time are more disruptive on material that is more similar"(Wickens, 2004, p. 131) the end of the verbal list becomes confusing. Because of the primacy and recency effects (Ritter, 2010, p.112) the first few and last couple of items to be remembered — which might be enough. Of course, as commands are learned through repetition (effectively encoding these commands into long term memory), this help list becomes a tool for recognition rather than recall. The user will only have to hear the command and identify what they need.

As a user I went in with several incorrect user models when using Sync Services: 1) a Interactive Voice Response (IVR) system, where you often have the ability to "cancel" or "go back" through voice or keypad entries, and 2) Google Maps, where the system will suggest things if it can't quite find what you mean. This caused proactive interference (Norman, 2008, p. 166)that made it difficult and frustrating to learn and use. Unlike the Phone or the Music player, Services has no obvious schema so I latched on to IVR but this proved to be counterproductive. In every operation there were no voice commands for canceling or "going back". As a result it was extremely difficult to hold a mental map of where I was in the process or even how to start over. I expected it to give me more choices or at least be suggestive but instead it just kept responding with choices that didn't make sense or were unhelpful.

Experience #2: Enrollment in myStanford connection

Scenario

For week 8 we had to register ourselves for the free MyStanfordConnection in order to watch a video by Jeff Johnson. The first thing that struck me was that the site was asking me if I was a new or returning user. Nothing abnormal there, except there was an asterisk (*) — and no reference point for this. I had no idea what it meant, but I scanned the page rapidly with top-down processing expectations to find another asterisk on the page. No such luck. This was confusing enough to momentarily pull my attention from the goal of signing up to this possibly insignificant mark on the page. I gave up and continued. After entering all my information in there was a checkbox that said "Create an Account". I wasn't sure what that was for, either, and the explanation made sense in itself, but the checkbox still did not.

After submitting information I was asked to enter a username and password, which I did and I hit return. The result was that I was just dumped to the home screen. I tried to log in, but I was given a screen that basically covered all the possible errors:

Shaken but not fazed, I tried the "forgotten password" process, which proved to be quite helpful:

Ultimately, the crux of this problem was the assumption that hitting return would properly submit the form, and I was able to create an account successfully after much head scratching. Regardless, this entire process itself was not helpful, for several reasons I will discuss.

Concepts for Analysis

Concept 1: Seven Stages of Action

The seven stages of action breakdown human goal seeking into a model sequence of steps. The seven steps are divided into three components of goals, execution and evaluation. The stages usually (but not always) begin with the goal— the desire to do something. This leads to first three steps: 1) an intention to act primed by the goal, 2) a cognitive choice of action 3) execution of that action. The result of the action is that there is some effect on the "world" — the environment, the system, or some external entity. The steps of evaluation follow: 4) perception of the "world", 5) interpretation or cognition on that perception, and finally, 6) a judgment or evaluation whether or not the intended goal was achieved.

This series of steps constitutes a theoretical model of the process; it is by no means a logical progression that always occurs in this order or without disruption. Often goals are formed after perception and cognition, called "Opportunistic actions" (Norman, p.48). Within the process there are gaps that impede the progress: The Gulf of Evaluation is a cognitive chasm that occurs when the environment or system fail to provide sufficient feedback for the person to interpret and verify the result of action. The Gulf of Execution occurs the environment or system do not provide actions relevant to the pursuit of the goal — or any actions at all.

Concept 2: Cognitive Styles

When designing a system or interface it's useful to consider the ways your potential users might expect to interact with it. A cognitive style "falls somewhere between a cognitive ability and a personality trait." (Norman, 2008, 242). Attending to these styles that differentiate users at best make a system more understandable, at worst at least make the process less frustrating and annoying. These styles tend to have polarizing traits. Some of cognitive styles relevant to my analysis:

Analytic vs. Holistic. Analytic individuals tend to want to delve into the finer details of a problem, picking apart the system as much as possible to solve an issue, whereas holistic thinkers tend to run with a more general, broader view, relying more on trial an error than planning and heuristics.

Reflective-Impulsive. A reflective user, like a chess player, will delve into multiple hypotheses about a situation and contemplate all the alternatives. Impulsives will instead go with one of the first choices.

Need for Closure. Akin to reflective/impulsive, those with a high need for closure require obtaining a definitive answer, although this is "done in a reasonable amount of time"(Norman, 2008, p.244), versus an exhaustive search. On the other hand those with a low need for closure will tend to accept ambiguity and results that do not provide a clear answer.

Analysis

The first confusing encounter with creating an account here was the "Create an Account' checkbox. My goal here was to simply make an account. This little checkbox caused a momentary small Gulf of Evaluation even before I had clicked submit. Wasn't that the purpose of entering all this information in the first place? I had to ask myself whether or not creating an account was actually what I wanted to do here, even though that was my original goal. The explanation was:

This sentence made sense, but the checkbox still didn't; it wasn't immediately clear why aI would enter all this information without that intention. I left it checked.

When I impulsively hit return after entering my password selection, I expected the form to be submitted— this is the behavior I expected from web forms, and so I had this conceptual model in my head. When I got nothing but a jump back to the home screen, this immediately created a Gulf of Evaluation. Had I successfully submitted the form? Here was a distinct lack of visibility as I had no idea what had just happened or more importantly, how to act or what actions were available.

Running with the assumption that I had in fact made an account, with or without feedback, I attempted to log in. This resulted in another error, and despite being given an elaborate feedback page indicating the possible errors this created yet another Gulf of Evaluation. What exactly was my error, and how could I correct it? The blanket statement given to me did not make it easy to get an answer here.

Being a user with a moderately high need for closure (partially driven here by the ultimate need to have access to this site), I was content with this answer. I also realized I was more Holisticand Impulsive in this situation, notwanting to pick apart why this might be happening and acting quickly on the options I found available. The first one was the larger second bullet which indicated I could recover a password. For all I knew, I just typed it wrong.

As the last screen shot indicates, however, the forgotten password link created a final Gulf of Evaluation as there was no system visibility— nowhere to enter any information to even try to recover the password. I did try submitting but of course that resulted in an error. As an available action, it was confusing and not sufficient in pursuit of my goal.

As stated in the scenario, I did finally realize that the problem was the form wasn't being submitted with a return keystroke. I complied with the system's constraints to actually click the submit button, and finally was able to achieve my intended goal… closure achieved.

Experience #3: Exposé across Mac OS X vs. Windows Flip 3D

Scenario

I started using Exposé and Spaces in Mac OS X 10.3. Generally speaking I found that it worked really well at its inception, although I had little insight as to exactly why it would be working well. As it turns out Exposé works because of the ways that it leverages our visual scanning and searching abilities — in a manner that succeeds over Windows Flip 3D. With OS 10.6 Apple introduced some changes to Exposé: windows now organize themselves in a grid, rather than placing them seemingly at random and text is used as a supplement to the search. It also reveals all the minimized windows, something that the original Exposé did not do, which could be a good or a bad thing.

Concepts for Analysis

Concept 1: Visual Scanning.

The act of perception begins the process of cognitive processing with input from our senses. Our eyes have a limited Useful Field of View, an "angle of no more than about 2 degrees surrounding the center of fixation" (Wickens, 2008, p.76). As such, in order to take in a sufficient amount of perceptual information to make any sense out of it requires eye movement. There are two types of eye movement when scanning: pursuit, where the eye tracks input in a smooth motion, and saccadic, where the eye jumps from location to location in order to focus on a destination or dwell (Wickens, 2008, 78). Saccadic movement is driven largely by the information content (the amount of information to take in across dwells) and time for information extraction (the objective quality of the information) (Wickens, 2008, 79). The two types of cognition that occur in the latter process are top-down — where active thinking intercepts perception to "guess" based on expectations and/or expertise, and bottom-up, where mere salience is enough to "grab" our attention.

Concept 2:Visual Search – target search.

How long it takes us to find a target involves several factors. The number of targets needed to search through, the salient dimensions of the target(s) sought, and whether or not targets are organized in anyway. With an unknown target location "on the average, the target will be encountered after halfof the targets have been inspected" [ibid]. Without sufficient salience or differences from other distractor targets [ibid], serial search "is more likely when the target is difficult to discriminate" (Geisler & Chou, 1995). With salience a search can be conducted in parallel instead, a more holistic manner that doesn't involve cognitively process every possibility in order to determine a target's properties.

Without organization of the target field, searches "tend to be considerably more random" [Wickens, 2004, p.80]. If the field of information is organized, users tend to tend to search top to bottom and left to right [ibid].

Analysis

The original implementation of Exposé would take the collection of visible windows on the Mac OS Desktop and place them around the screen at proportional sizes:

The windows are also grouped by application. Exposé uses a (theoretically) smooth visual transition to this alternative view, making use of the Wicken's (2008) bottom-up process conspicuity to draw attention to itself. Depending on the system, the ability to use pursuit scanning varies as the window transition. Generally the movement is too quick or jerky to facilitate this kind of tracking, although Exposé does allow the option to significantly slow down the transition to aid this kind of tracking.

When compared to the task of just app switching alone, this process has several advantages. With many windows on screen, app switching merely brings the active group of windows to the foreground without any movement to attract attention or initiate any pursuit scanning. First, simple lack of occlusion lets each window be seen in its entirely, including movement if the window happens to be a movie or animated object. Each window is allowed to rely on its own inherent visual appearance for salience; Apple hopes that users will be able to identify the window they want by its contents. In the above example, the purple window or the chess game are both visually salient in context and will be easy for the user to find.