Hick's law

Hick's law, also called the Hick-Hyman law, describes how reaction time increases as the number of possible choices increases. It comes from two lab studies: W. E. Hick's 1952 paper "On the rate of gain of information" and Ray Hyman's 1953 paper "Stimulus information as a determinant of reaction time". For equally likely options, the usual form is T = b · log2(n + 1), where n is the number of choices and b is a constant measured empirically (summary on Wikipedia).

Why it matters

The law is a reminder that every extra option has a cost, even if it is small. Each option is something to perceive, compare and rule out. In time-critical moments (a checkout button, an alert that needs a response, a confirm dialog) a short, clear set of choices helps people act quickly.

The logarithmic shape matters too. Going from 2 to 4 options costs about as much extra time as going from 8 to 16. So the law does not say that long lists are unusable. NN/g's video on long menus makes the point that "Combining Hick's Law with other design techniques can make long menus easy to use."

How to apply it

• Do reduce options where speed matters. A dialog that needs a fast answer should have one obvious primary action.
• Do highlight a recommended option when most people want the same thing, such as a default plan or the most common export format.
• Do group and order long lists (alphabetically, by category, by frequency) so people can narrow down instead of reading every item.
• Do break complex flows into steps, and reveal advanced settings with progressive disclosure.
• Don't cut options people genuinely need just to hit a small number. Laws of UX warns: "Be careful not to simplify to the point of abstraction."

Common mistakes

• Using Hick's law to explain complex decisions. The original studies measured choice reaction time to simple stimuli (which light came on, which key to press). Picking a pricing plan or a holiday involves reading, comparing and judging, which the law does not model.
• Ignoring its assumptions. The formula holds for equally likely options. When some choices are far more common, or people already know what they want, decision time does not follow the simple count of items.
• Applying it to unordered lists. Wikipedia notes that finding a word in a randomly ordered menu requires scanning each item, which takes linear time, so Hick's law does not apply. Ordering the list is what makes faster search possible.
• Treating it as a reason to hide everything. Too few top-level choices can create deep menus that are slower overall.

In AI products

AI interfaces often stack choices: a model picker, a tone selector, a row of prompt suggestions, and several actions on each response. A handful of well-chosen suggestion chips helps people start. A dozen generic ones becomes another list to read. Pick a sensible default model, show a few suggestions tied to the user's context, and keep rarely used controls one step away.