Monday, October 12, 2009

Human Capabilities in HCI

Human sensation, perception, cognition, and action are all very tightly integrated. Traditionally, experts have separated the studies of these systems into an input class which encompasses both sensation and perception, and an output class consisting of studying behavior and motor expression, with a space in between for processing- akin to the black box model. However I would argue that they are so tightly coupled that it is impossible to draw a distinct line which separates them succinctly. Rather, these systems- only a small portion of which can be physically observed to the naked eye- flow smoothly into each other, reminding us how interdependent they really are. Perhaps the most important facet of these couplings is the feedback loop that exists between the input and output classes of study, and how that relationship both affects and is affected by the more cognitive systems like attention and memory. More specifically, these systems and their relationships play a vital role in how we go about creating technology in a symbiotic manner.

The paper "Perceptual-Motor Interaction: Some Implications for HCI" contributed by T. Welsh, R. Chua, D. Weeks, and D. Goodman describe two theoretical and analytical frameworks for studying perceptual-motor interaction in HCI. In the first framework, the authors take the view of perceptual-motor interaction in the context of an information processing model. In this, they point to (pun unapologetically intended) how humans currently issue commands and select objects in user interfaces, traditionally using a mouse. They mentions Fitts' Law- a predictive model of time to engage a target- which essentially states that targets are easier to acquire if they are closer and bigger, and that the difficulty of accurately attaining a target is a function of the distance and width relative to the current position. Designers should be aware that commonly expected actions need to larger in diameter and should be as close to the starting point as possible. Another design implicative fallout from Fitts' law is the 5 fastest locations that a mouse cursor can access at any given time (assuming the cursor is at the center to start with, these locations are: the center, and the 4 corners- since each corner has infinite target width in both the x and y axes). Some interfaces have taken this to heart (Google Chrome's tabbed browser features tabs that reach all the way to the top edge of the screen), and many have not. It should be noted that Fitts' Law has successfully been applied to direct manipulation interface- not only in the case of mapped interfaces requiring mouse interaction.

The authors also mention the Hick-Hyman law, which predicts the decision time required to select a target response from a set of potential responses. This law states that the amount of time it takes to make a choice increases with the number of possible choices. This has direct implications for design in that task completion time increases greatly when you provide more menu options, more buttons, more labels, etc. This is a common argument against scope creep. This problem is not only applicable to Human Computer Interaction, however. In this TED talk, Barry Schwartz describes the paradox of choice, and how even though it may seem to follow that we increase satisfaction by offering users more options, all it does is make it take longer to choose (inducing chooser's paralysis), and ends up making them feel worse afterwards (due to chooser's regret).

The paper goes on to describe Attention, and how it affects perceptual-motor interaction due to its limiting affect on one's capacity to process information. The authors offer three important characteristics of attention:
  1. Attention is selective and allows only a specific subset of information to be processed
  2. Focus of attention can be shifted from one source of information to another
  3. Attention can be divided such that, within certain limitations, one may selectively attend to more than one at a time (the cocktail party phenomenon)

Attention is closely associated with ethics in web design now, since many companies have learned how to manipulate users' attentional characteristics to get them to pay attention to their advertisement on web sites. For instance, using movement, color, and other stimuli which break that barrier discussed in the divided attention (cocktail party) model, companies can lasso users' attention away from their original task material and toward their offers. Perhaps what web-savvy marketing companies didn't count on was a phenomenon known as "Banner Blindness." Essentially, users have rather quickly adapted to learn how advertisements look and behave on websites to the extent that their attention has learned to block it out. In accordance with the first described characteristic of attention, users' began to selectively ignore banner ads on websites. Lately, advertisers have had to become more clever in capturing the attention of website visitors, employing tactics like hidden ads, interactive ads, and fly-across ads.

The paper wraps up the discussion on perceptual-motor interaction in applied tasks by illustrating a few examples: remote endoscopic surgery, PDA operation, and eye-gaze vs mouse interactions. The last example exemplifies the power that eye gaze technologies may have in future selection activities, citing that they may be more efficicent that those of the traditional manual (mouse) systems. While supported by new and upcoming research activities, I would like to suggest caution in assuming superiority in every day actions with use of eye gaze interfaces. While there are several critiques, one that stands out to me is that of the Midas Touch problem. Human movements are easy, voluntary, and natural. Conversely, eye gaze commands are not necessarily natural for interface commands, not easy to do with high accuracy, and most importantly, are not always voluntary. Unlike a simple purposeful twist of a knob or a flick of a switch, the eyes are always "ON." How is an interface to know when to accept gaze as input, and when the user is just looking? This problem is a tricky one and needs to be addressed in any interface system where false-positives are likely to be an issue (gestural interfaces, Brain Computer Interfaces).

The eye gaze vs. mouse investigation reveals the importance and difficulty in HCI around trying to optimize for user productivity, in so far as finding ways to make interactions more efficient by taking advantage of how humans can interaction in particular ways. In the following concept video, the makers make a case for an alternative interface. The attempt here is to optimize our ability to interact with our fingers, while at the same time optimize for ergonomic comfort and visibility- common critiques of current touchscreen technologies.

10/GUI from C. Miller on Vimeo.

No comments: