Gestalt psychology (Wertheimer 1923 and others, in the first half of the 20th century) discovered that people don't look at a scene pixel by pixel — they subconsciously organize the parts into a structured whole: 'the perception of the whole is not the sum of its parts.'

These 'grouping principles' directly determine, in an interface, 'what looks like a group and what looks separate.' The original set includes proximity, similarity, closure, and continuation; late in the 20th century common region (Palmer, 1992) was added, among others.

What it means for designers: you don't need a label saying 'these belong together' — place and style elements using Gestalt cues, and users naturally read them as a group.

Proximity: sits close = one group. This is the most effective grouping cue and the one to reach for first. Elements near each other are perceived as belonging to the same group; distant elements are read as playing different roles.

How strong is it? Often a simple spacing adjustment beats adding divider lines or boxes.

In practice: pull each form label close to its own input field, and leave larger gaps between groups — users instantly read the structure. The anti-pattern is a label sitting closer to the previous input than to its own — users then type into the wrong row. Group with spacing first; bring in other means only if that's not enough.