Cognitive Load and the Limits of Hick's Law
Why fewer options usually helps, and why the famous law does not prove it.
Fernway, a travel expense tool, has a screen called New Expense. It opens with a currency dropdown holding 168 entries, a category picker with 31 options, a project field, a cost centre field, and a toggle labelled Billable. Support sees the same ticket every week: people submit expenses in the wrong currency. A designer proposes cutting the currency list to five. Someone on the call says Hick's law. Everyone nods. Nobody checks what Hick's law actually says.
The cut is probably right. The reason given for it is wrong, and that matters, because the wrong reason will get applied to the next screen where it does not hold.
Where the difficulty actually sits
What Hick actually measured
W. E. Hick published his result in 1952, and Ray Hyman published a companion result in 1953. Participants sat in front of a small set of lamps, each paired with a key. A lamp lit, they pressed the matching key, and the experimenters timed them. Reaction time rose with the logarithm of the number of alternatives. Doubling the lamps did not double the time, it added a roughly constant increment.
Read the setup again. The responses were learned in advance, the alternatives were equally likely, and there was nothing to read or interpret. None of that describes a currency dropdown. Picking USD from a sorted list is a visual search problem plus a reading problem, and a sorted list of 168 items with a type-ahead can be faster than an unsorted list of 12. The log relationship does not survive the trip from lamps to labels.
Three costs hiding under the word complexity
- Search cost
- How long it takes to find the option you already know you want. Sorting, grouping, and type-ahead cut this. Deleting options barely helps if the list is well ordered.
- Decision cost
- How long it takes to work out which option you want. This is where a long list genuinely hurts, because every option is a small comparison you have to run.
- Working memory cost
- How much you have to hold in your head while you decide. Sweller's cognitive load: How much a person has to hold in mind at once to get through a task. theory is the useful frame here, and it points at the instructions and the interdependencies, not the raw option count.
Fernway's real problem is decision cost sitting where it does not belong. A traveller on a trip to Lisbon knows the currency. They do not need to decide, they need the field to already say EUR. The fix is a default from the receipt or the trip, with a change link for the rare case, and that fix has nothing to do with list length.
Worked example
The category picker, retold
Fernway's 31 categories exist because finance wants clean data. Users want to file the receipt and leave. The team keeps all 31 for the finance export but shows six on the form: Flights, Hotel, Meals, Ground transport, Software, Other. Choosing Other reveals the full list. Miscategorisation drops because most people never meet the long tail, and finance still gets its taxonomy. Nothing was deleted.
You will also hear that people can hold seven items in mind, from George Miller's 1956 paper. Miller himself was careful about it, and later work, including Nelson Cowan's review, puts the practical limit closer to four for items you cannot rehearse. Neither number is a design rule. Both get quoted as one.
It doesn't matter how many times I have to click, as long as each click is a mindless, unambiguous choice.
Fields that force a decision
- Project
- Category
- Currency
- Billable
The same fields, answerable
- Project: Lisbon site visit (from your trip). Change
- Category: Meals. Detected from the receipt.
- Currency: EUR. Change if you paid in another currency.
- Bill this to the client? Your last 6 Lisbon expenses were billed.
Krug's line is the better instrument. Count the ambiguous choices on a screen, not the controls. A form with 20 fields where every answer is obvious is easier than a form with four fields where one asks you to guess what the team means by Project. When someone in a review says a screen has too much on it, ask them which specific choice they could not make, and fix that one.
Quick check
Why is Hick's law a weak justification for shortening a labelled dropdown?
The takeaway
Option count is a weak proxy for difficulty; the real cost is the number of choices a user cannot answer without thinking.
Try this tomorrow
Open your highest-traffic form and mark every field a user cannot answer in under two seconds. Fix the worst one with a default or a clearer label before you delete anything.
Answer the check above, then bank the day.
Where this comes from
- On the rate of gain of information (1952), W. E. Hick
- Stimulus information as a determinant of reaction time (1953), Ray Hyman
- Cognitive load theory, John Sweller
- Don't Make Me Think, Steve Krug
User Psychology is one of six tracks. These lessons summarise and build on the work above, they do not reproduce it. Buy the books, they are better.