The Simon effect: when where a thing is fights what it says

What it is

J. R. Simon and A. P. Rudell reported the effect in the Journal of Applied Psychology in 1967. Participants heard the spoken words “left” and “right” presented randomly to one ear or the other, and responded according to the word. Which ear delivered the word carried no information at all — it was irrelevant by construction, and participants were told so. Responses were nevertheless slower when the word arrived at the ear opposite the response it called for.

That is the whole finding, and its importance is in what it rules out. Nobody had to interpret the ear, remember it, or resolve it against anything. The irrelevant spatial dimension was processed regardless, and it activated the response that spatially corresponded to it. When that activation matched the correct response, it helped; when it did not, it had to be overcome.

The visual version most people meet today is a later adaptation: a coloured shape appears left or right of centre, the participant presses a left or right key according to colour, and the position of the shape is irrelevant. Responses are faster on congruent trials — stimulus and response on the same side — than on incongruent ones. Bernhard Hommel’s 2011 review in Acta Psychologica traces how a curiosity from human-factors research became a standard instrument for studying how actions are represented and controlled.

The effect belongs to a wider phenomenon that Paul Fitts and C. M. Seeger had already established in 1953: stimulus–response compatibility. They showed that some pairings of a stimulus set with a response set are simply better than others, and that the cost of a poor pairing survives practice. The Simon effect is the sharper case, because in Fitts and Seeger’s work the spatial dimension was the thing being responded to, whereas in Simon’s it was explicitly not.

The level in this game is a Simon task rather than an illustration of one. An arrow appears on the left or right half of the screen and points left or right; the two are randomised independently, so roughly half of all trials are incongruent. You respond to the direction the arrow points, and the side it appeared on is irrelevant. You need eight correct in a row, and a single wrong answer resets the count to zero — which is what makes an effect worth a few tens of milliseconds and a few per cent of errors something you can actually feel.

Why it works on people

Spatial location is encoded whether or not it is wanted. The visual and auditory systems deliver “what” and “where” together, and there is no setting that suppresses the second. Instructing someone that position is irrelevant does not stop position being represented; it only tells them not to use it deliberately.

Once position is represented, it primes the action that corresponds to it. The dominant account holds that a stimulus on the left partially activates a left-hand response automatically, in parallel with the slower, deliberate route that works out the correct response from the relevant feature. On congruent trials the two routes agree and the response is quick. On incongruent trials they disagree, the automatic activation has to be suppressed, and both time and accuracy suffer.

This is a different failure from the Stroop effect, and the distinction matters for diagnosis. In Stroop, the conflict is between two features of the same object — a word and its ink colour — and it is resolved at the level of meaning. In the Simon task, the conflicting dimension is not a property of the object at all but of where the object is, and the conflict is resolved at the level of action. A design can be perfectly clear in its wording and still lose to the Simon effect, because the wording is not what is competing.

The cost is small per event and reliably present. That combination is what makes it a production problem rather than a usability-session problem: a moderator watching six participants perform a task twice will see nothing, while the same design across a million interactions produces a steady rate of wrong-side clicks that nobody can reproduce on demand and everybody blames on carelessness.

Practice reduces the effect but does not remove it, and it returns whenever attention is divided or the user is hurried. An expert user of your product is not immune; they are simply operating with a smaller margin, which is exactly when a competing automatic response is most likely to win.

Where you meet it

• A confirmation dialog whose “Yes” sits on the right while the thing it will delete is highlighted on the left of the screen behind it.
• A split-pane settings screen where the control column is on the right and the preview it modifies is on the left, so every adjustment is made in the direction opposite the feedback.
• Notification toasts that appear in one corner and whose action button is in the opposite corner of the screen, in a persistent bar.
• Media and carousel controls placed to the left of the content they advance to the right, or transport controls mirrored between a mini player and its expanded view.
• “Previous” and “Next” buttons positioned in the reverse of their reading order, most often when a right-to-left layout is produced by mirroring the container but not re-checking the semantics of each control.
• A list where each row’s action button sits on the far side of the row from the item’s identifying label, so the eye lands on the left and the hand must commit on the right.
• Keyboard and gamepad mappings where a directional input produces movement in the opposite direction, which is the same conflict expressed through the motor system rather than the visual one.
• Drag-to-reveal row actions that place a destructive action on the side the user swipes from, so the gesture direction and the action’s position disagree.

Designing around it

• Put the control on the same side as the thing it acts on. This is the whole remedy, it is nearly free, and it is more effective than any amount of labelling — because the conflict is not about meaning and labels do not address it.
• Keep spatial mappings consistent across the entire product, not merely within a screen. The automatic response is built from experience, so a single screen that mirrors the convention everywhere else establishes competing expectations and makes every screen slightly worse.
• Where a control must be separated from its object, bind them visually so the pairing is unambiguous: a connecting line, a shared container, a highlight that fires on hover or focus. You cannot remove the spatial code, but you can make the correspondence obvious enough that the deliberate route wins comfortably.
• Treat right-to-left layouts as a semantic problem rather than a CSS one. Mirroring a layout moves the spatial correspondences too, and controls whose meaning is intrinsically directional — a media seek bar, a chart axis, a real-world orientation — may need to stay put while the rest of the interface flips. Decide this per control, deliberately.
• Do not use position alone to distinguish two actions with different consequences. Position is the dimension most vulnerable to automatic capture, so an irreversible action separated from a benign one only by which side it sits on is the worst possible arrangement. Separate them by region, by control group, or by requiring a different gesture.
• Watch for the Simon effect specifically when adding a second view of the same controls. Mini players, compact modes, condensed tables and responsive breakpoints are where mirrored mappings are introduced, because each was designed on its own and the spatial relationship between them was never a review item.
• Measure it with congruent and incongruent variants rather than by asking. Users cannot introspect on a fifty-millisecond conflict and will report that both layouts feel fine. Build both, run the same task repeatedly under mild time pressure, and compare error rates rather than opinions.
• Instrument rapid corrections. A click followed within a second or two by an undo, a back navigation, or the opposite action is the signature of a response that lost to the wrong automatic one, and that sequence is usually already in your event stream.

Questions

What is the Simon effect?

It is the finding that people respond more slowly and less accurately when the position of a stimulus does not correspond to the position of the required response, even when position is irrelevant to the task. Simon and Rudell demonstrated it in 1967 by playing the words “left” and “right” into one ear or the other; responses were slower when the ear and the answer disagreed, despite the ear carrying no information.

How is the Simon effect different from the Stroop effect?

The conflicting dimension is different, and so is the level at which it is resolved. Stroop pits two features of the same object against each other — the word and the ink it is printed in — and the conflict is about meaning. The Simon effect pits an object’s location against the response it calls for, and the conflict is about action. This is why clearer wording fixes a Stroop problem but does nothing for a Simon one.

Does the Simon effect matter if my labels are clear?

Yes, because labels are not what is competing. The effect arises from a control’s position automatically priming the spatially corresponding action, which happens in parallel with reading. A perfectly worded button on the wrong side of the screen still collects wrong clicks, and the users making them will not be able to tell you why.

Can users learn their way out of it?

Only partly. Practice shrinks the effect but does not eliminate it, and it re-emerges under time pressure, divided attention or fatigue. Expert users are not protected — they work faster, which is precisely the condition in which an automatic response is most likely to beat a deliberate one.

What is the single cheapest fix?

Put the control on the same side as the thing it affects, and keep that mapping identical everywhere in the product. Spatial correspondence costs nothing to implement, addresses the actual mechanism rather than a proxy for it, and compounds in value because the automatic response is built from repeated exposure to your own conventions.

Related patterns

• The Stroop Effect: when a label and its colour disagree
• The flanker effect: when neighbouring controls compete for your response
• Signifiers: an affordance nobody can see is not a feature