- Explain selective and divided attention and their limits
- Describe Hick's Law and its implications for menu and option design
- Understand how cognitive biases affect user decision-making
- Apply principles of choice architecture to interface design
- Identify strategies for managing interruptions and maintaining task focus
Introduction
Perception delivers information; memory holds it; but attention determines which information is processed and which is ignored. The human attentional system is fundamentally a bottleneck — there is far more information available to the senses at any moment than the brain can process. How that bottleneck is managed, and how it interacts with decision-making, determines whether a user completes a task efficiently, makes an error, or abandons the task altogether.
Selective Attention
Selective attention is the ability to focus on one source of information while ignoring others. Broadbent's filter theory Broadbent, 1958 proposed that attention acts as an early filter, selecting one channel of information for processing and blocking others. Later research by Treisman and Deutsch showed that unattended information is not completely blocked but is processed to a shallow level — enough to detect one's own name in an unattended conversation (the "cocktail party effect") but not enough for detailed comprehension Treisman, 1980.
Visual Search
Much of interface use involves visual search: finding a specific button, menu item, or piece of information in a display. Visual search can be either parallel (when the target differs from distractors in a single pre-attentive feature) or serial (when the target is defined by a conjunction of features or is visually similar to distractors).
The time required for visual search depends on how easily the target can be distinguished from its surroundings. Targets that differ in a single pre-attentive feature (colour, size, orientation) pop out regardless of the number of distractors. Targets that share features with distractors require serial scanning, and search time increases linearly with the number of items. Interface design should make frequently used or critical targets visually distinctive.
Implications for Interface Design
Menus with many visually similar items — long lists of similarly formatted text entries — require serial search. Techniques that reduce search time include:
- Grouping related items (using spatial separation or visual dividers)
- Highlighting the most likely selection
- Ordering items by frequency of use or logical category
- Limiting the number of items visible at once (progressive disclosure)
Divided Attention
Divided attention — attempting to attend to multiple tasks or information sources simultaneously — is severely limited. Dual-task performance suffers whenever both tasks require the same cognitive resources Wickens, 2008.
The Psychological Refractory Period
When two tasks requiring responses arrive in rapid succession, the response to the second task is delayed. This psychological refractory period (PRP) reflects a central bottleneck in response selection. In interface design, this means that rapid-fire notifications or simultaneous demands for input will cause delays and errors.
Multitasking and Task Switching
True multitasking — simultaneous processing of two demanding tasks — is largely a myth for tasks requiring central cognitive resources. What people call multitasking is typically rapid task switching. Each switch incurs a time cost (the switch cost) of 200–500 milliseconds and an accuracy cost, as the mental context of the previous task must be suppressed and the new task's context loaded into working memory.
Westbrook and colleagues Westbrook, 2010 observed hospital nurses preparing and administering medications and found that interruptions were common and measurably harmful. Each interruption was associated with a 12.1% increase in procedural failures and a 12.7% increase in clinical errors. The interruptions forced task switching — the nurse had to suppress the medication task, deal with the interruption, then reload the medication task context into working memory, with each switch creating an opportunity for error. The study is one of the strongest empirical demonstrations that divided attention in safety-critical work has a direct, quantifiable cost.
Interruptions and Resumption
When a user is interrupted during a task, they must later resume it. Research on interruption and resumption shows that:
- Longer interruptions lead to more errors on the resumed task
- Interruptions at task boundaries (between subtasks) are less disruptive than mid-subtask interruptions
- Environmental cues that preserve the state of the interrupted task (a visible trail of completed steps, an autosaved draft) reduce resumption errors
Design for interruption. Assume that users will be interrupted and must resume later. Preserve task state automatically. Provide clear indicators of where the user left off. Allow tasks to be saved and resumed at any point, not just at designated checkpoints.
Hick's Law
Hick's Law (also called the Hick-Hyman Law, after W. E. Hick Hick, 1952 and Ray Hyman Hyman, 1953) describes the relationship between the number of choices and the time required to make a decision. RT = a + b × log2(n) where RT is reaction time, n is the number of equally probable alternatives, and a and b are empirically determined constants. The logarithmic relationship means that doubling the number of choices does not double the decision time — it adds a constant increment.
Hick's Law predicts that reaction time increases logarithmically with the number of choices. A menu with 8 items does not take twice as long to decide from as a menu with 4 items — it takes only about one "unit" longer. However, the law applies strictly to equally probable, equally familiar choices. In practice, frequently used items, distinctive items, and items the user is specifically looking for may be selected much faster than Hick's Law predicts.
Implications and Limitations
Hick's Law is sometimes misapplied to argue that menus should always have fewer items. This is misleading for several reasons:
- The law applies to decision time, not search time. If the user must search through items to find the right one, search time (which may be linear in the number of items) dominates. Hick's Law describes the decision among known alternatives.
- Unequal probabilities change the calculation. If one option is much more likely than others, the effective number of choices is reduced. Placing the most common action prominently can reduce decision time more than removing less common options.
- Hierarchical menus trade decisions for depth. Splitting 64 items into 8 groups of 8 reduces each individual decision but requires two decisions in sequence. The total time depends on the specific structure and the user's familiarity with it.
Consider a restaurant menu with 100 items. Hick's Law predicts a decision time proportional to log2(100) ≈ 6.6 "units." But in practice, diners do not consider all 100 items equally. They first decide on a category (starter, main, dessert), then choose within it. The hierarchical structure reduces the effective number of choices at each stage. When does categorisation help, and when does it add unnecessary navigation?
Decision-Making and Cognitive Biases
Human decision-making systematically deviates from rational optimisation. Kahneman and Tversky's research on heuristics and biases [Tversky, 1974; Kahneman, 2011] identified numerous ways in which decisions are influenced by factors that should, rationally, be irrelevant.
Anchoring
The first piece of information encountered exerts a disproportionate influence on subsequent judgments. A default value in a form field acts as an anchor — users are more likely to accept values close to the default even when a different value would be more appropriate. This makes default selection a powerful and responsibility-laden design decision.
Framing Effects
The way options are described affects choice. "95% success rate" and "5% failure rate" are logically identical but produce different decisions. In interface design, the wording of options, error messages, and confirmations can frame decisions and influence user behaviour.
The Default Effect
Users disproportionately accept default options — whether through active preference, satisficing (accepting a "good enough" option to reduce effort), or inattention. Research on organ donation rates across countries shows dramatic differences explained largely by whether the default is opt-in or opt-out.
Defaults are among the most powerful design decisions. Because users overwhelmingly accept defaults, the designer who chooses the default value is effectively choosing the outcome for most users. Defaults should be set to the option that is safest, most beneficial, or most commonly desired. In safety-critical systems, defaults should always fail safe.
Satisficing vs. Optimising
Herbert Simon's concept of satisficing Simon, 1956 — choosing the first option that meets a minimum threshold rather than evaluating all options to find the best one — describes how users typically interact with interfaces. Users do not read every menu item; they scan until they find something that looks right and select it Krug, 2014. This behaviour is efficient but error-prone when multiple options look similar.
Choice Overload
While more options might seem better, research by Iyengar and Lepper Iyengar, 2000 demonstrated that too many choices can lead to decision paralysis, reduced satisfaction, and avoidance. Their jam study found that shoppers presented with 24 varieties were less likely to purchase than those presented with 6.
Streaming services face the choice overload problem directly. Netflix's recommendation system is, in part, a response to the observation that users presented with thousands of titles often browse without watching. The recommendation algorithm reduces the effective number of choices by filtering and ranking options based on predicted preference.
Choice Architecture
Choice architecture, a term popularised by Thaler and Sunstein Thaler, 2008, describes how the structure of a decision environment influences choices. Key principles include. Structuring complex choices. When many options exist, provide categories, filters, or sorting mechanisms that allow users to reduce the choice set incrementally. Making consequences visible. Users make better decisions when they can see the likely outcome of each option. Preview features, simulation modes, and clear labelling of consequences all support informed choice. Providing feedback. Immediate feedback on decisions — showing the result of a configuration change in real time, for instance — allows users to evaluate and revise choices before committing. Expecting error. Designs that anticipate common decision errors — confirmation dialogs for destructive actions, undo functionality, and clear warnings about irreversible choices — reduce the cost of mistakes.
Attention in Monitoring Tasks
Some tasks — clinical monitoring, air traffic control, security surveillance — require sustained attention to detect infrequent events. The human attentional system is poorly suited to these vigilance tasks. Mackworth's clock test Mackworth, 1948 demonstrated that detection rates for infrequent signals decline within 30 minutes.
Do not rely on sustained human vigilance for detecting rare events. After approximately 20–30 minutes, human detection rates for infrequent signals decline significantly (the vigilance decrement). Automated monitoring with active alerting is more reliable for rare-event detection. Reserve human attention for interpreting and responding to alerts, not for detecting them.
Key Takeaways
- Selective attention is a bottleneck: the brain cannot process all available information simultaneously, and visual search time depends on target-distractor similarity.
- Divided attention and multitasking incur significant performance costs; each task switch carries time and accuracy penalties.
- Hick's Law predicts that decision time increases logarithmically with the number of equally probable choices, but search time, familiarity, and item distinctiveness modify the practical impact.
- Cognitive biases (anchoring, framing, the default effect, satisficing) systematically influence user decisions in ways designers must anticipate.
- Defaults are powerful: most users accept them, making default selection a high-impact design decision.
- Choice overload can impair decision-making; structuring and filtering options helps.
- Sustained vigilance degrades after 20–30 minutes; automated alerting is more reliable for detecting rare events.
Further Reading
- Hick, W. E. (1952). On the rate of gain of information. Quarterly Journal of Experimental Psychology, 4(1), 11–26.
- Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.
- Thaler, R. H., & Sunstein, C. R. (2008). Nudge: Improving Decisions About Health, Wealth, and Happiness. Yale University Press.
- Iyengar, S. S., & Lepper, M. R. (2000). When choice is demotivating: Can one desire too much of a good thing? Journal of Personality and Social Psychology, 79(6), 995–1006.
- Westbrook, J. I., et al. (2010). Association of interruptions with an increased risk and severity of medication administration errors. Archives of Internal Medicine, 170(8), 683–690.