People
Human-Computer Interaction
03 Sep 2026
Week TWO
Part II of Hornbæk et al. (2025)
Perception · Motor Control · Cognition
Intro
You can’t design for people you don’t understand. It’s tempting to start from how you want people to act, but people bring their existing perception, motor habits, and mental models to everything they touch, and those systems don’t renegotiate on your schedule.
This week’s question: what does the human processing pipeline (seeing, moving, thinking) let designers predict before anything ships?
Case Study: Healthcare.gov
What went wrong with the Healthcare.gov rollout?
Aside from the obvious (not anticipating the volume)
- They required users to create an account before browsing any plans, so the registration system bottlenecked under the load of curious browsers who had no way to evaluate the product first
- They underestimated the proportion of users who were “uncommitted shoppers” wanting to compare options before investing effort: a basic mismatch between system functionality and how people actually behave when shopping for insurance
- The designers’ model was citizens completing enrollment; the users’ model was shopping site. The user’s model always shows up first.
Perception
Perception is not a camera
- The naive view: the eye is a camera, the display is the photo, perception is passive playback.
- The chapter’s view: perception actively constructs a percept from three ingredients.
- Sensory information: the raw transduction of light, sound, and contact into neural events.
- Expectation: priors from years of experience that fill in the blanks the senses leave.
- Attention: the strategic choice of what slice of the scene to process right now.
- Consequence: the same pixels produce different experiences in different users. Design does not fully determine what people see.
The three ingredients of a percept
Note
Perception is the ability to collect and organize information about the environment through physiological sensory systems. In HCI it is understood through three interacting components:
- sensory information (what the senses transduce)
- expectation (priors from prior experience)
- attention (the strategic sampling of the scene)
Why a constructed percept changes your design
- A UI communicates only through perception: it is the entire channel from machine to user. Get it wrong and nothing else matters.
- Experts see differently from novices: experienced users lean on long-term memory and expectation; novices are driven by raw visual features. Your design meets both.
- Prior products train your users: people arrive expecting the back button top-left, the primary action bottom-right. Fighting those priors costs you.
- Perception is active: users move, explore, and re-strategize. Norman’s perceived affordance lives here: the question is not “does it afford clicking” but “does the user perceive that clicking here is useful?”
- Design rule: design for the percept users will construct, not the pixels you shipped.
Visual saliency
Note
Visual saliency is the probability that a graphical element attracts visual attention during the first few seconds of viewing a display. It depends on the element’s visual features relative to the rest of the display, on user expectations, and on the user’s attentional strategy.
Spending your saliency budget
- Uniqueness wins attention: an element pops out when it is unique in a visual primitive: color, size, shape, orientation, or motion.
- Feature search vs conjunctive search: if the target is unique in one feature, users find it in under 100 ms, peripherally. If it needs a combination of features, they must fixate item by item, which is slow (Treisman & Gelade).
- Clutter is saliency’s evil twin (Rosenholtz): clutter is the state where everything tries to be salient, so nothing is. Congest every feature and you destroy the ability to guide the eye.
- The Post-it test: if you cannot pick a color, size, or shape that would make a new note stand out on the screen, the screen is cluttered.
- Design rule: spend your saliency budget. Make the one thing that matters unique; make everything else recede.
Grouping: the Gestalt principles
- Proximity: elements close together read as a group.
- Common area: elements inside the same enclosed region read as a group.
- Similarity: elements alike in color, size, or orientation read as a group.
- Continuation: elements along a continuous flow read as a group.
- Payoff (Brumby & Zhuang): visual grouping speeds menu search, but only with a few larger groups. Many tiny groups dilute the effect.
The hard limits of the visual system
Note
Ware’s windows of visibility name the physiological limits every display fights:
- Spectrum and color: we see roughly 380 to 780 nm, through just three cone types (trichromacy)
- Field of view: about 190 degrees horizontal, 125 vertical
- Contrast: the contrast sensitivity function sets which differences we can even detect
- Foveal vs peripheral: sharp vision covers only about 2 degrees, a thumbnail at arm’s length; everything else drops off fast
Design for color vision deficiency
- It is common: red-green deficiency hits about 8% of men of European descent (4 to 6.5% of Chinese and Japanese men), versus 0.4% of women.
- What breaks: affected users can’t reliably separate green from red, or blue from yellow, and there is no cure.
- The fix is design, not treatment: never encode meaning in color alone; add shape, text, or position as a redundant channel.
- Periphery compounds it: peripheral vision keeps color and shape but loses detail, so a tiny colored status dot in a corner is easy to miss.
- Design rule: if stripping all color would make the screen unusable, you leaned on color too hard.
In the wild