The placement of the mental preparation unit is governed by heuristic rules and is the main source of estimation error

What it is

The mental operator (M) represents a user's brief cognitive pause before starting a method, switching operator type, checking feedback, or applying a selection rule. It cannot be mechanically derived from the sequence; its placement depends on heuristic rules, and inserting a few too many or too few Ms shifts the total time substantially. This card and the card on insertion rules under "estimate by summing standard operator units" ask the same underlying question from different angles: that one explains, from the cognitive phenomenon of chunking, why a rule is needed at all; this one spells out what the rule actually says — the original KLM methodology did not hand down a vague "use judgment," but a set of heuristic rules that can be applied step by step.

Why it happens

The insertion rules from the early KLM methodology break down into several directly applicable criteria: by default, insert an M before every keystroke that does not belong to the same continuous input string, and before every pointing action that opens a new cognitive unit; delete that M if the following operation is fully anticipated — a skilled user already knows in advance exactly what comes next and needs no time to decide; if a string of "mental preparation plus keystrokes" belongs as a whole to one cognitive unit (typing a complete word or a fixed-format code), keep only the M at the very front of that string and delete the rest; and if a mental preparation is immediately followed by a physical action that has become fully automatic and needs no cognitive involvement (pressing a confirm key as a muscle-memory-level closing action), that M can be omitted too. Together, these rules turn "whether to insert an M" from a vague matter of experience into a procedure that different people can apply repeatedly and reach similar results — but even so, the rules still require the analyst to first judge "is this string of operations one cognitive unit" and "is this action fully anticipated for the user," and both judgments still carry a subjective component. That is exactly why this is the main source of estimation error: given the same operation sequence, two people both familiar with the rules can still judge "is this one cognitive unit" differently and end up placing the M in different spots.

Where it stops holding

This rule set was distilled from a particular era and a particular type of task — mainly text editing and command input — and there is no single correct insertion scheme across different user strategies; a high-frequency expert may, through repeated practice, have turned a step that once required judgment into something fully automatic, warranting more Ms deleted, while the same task for a skilled-but-not-quite-expert user who has still had real training may retain exactly the pauses the rules originally suggested keeping. The duration of the M constant itself also fluctuates with task pressure and familiarity, and is not a fixed value that transfers everywhere. Precisely because the rules themselves leave room for subjective judgment, no single KLM analysis should report just one precise number — it should run a sensitivity analysis showing how the result changes if the insertion judgment shifts slightly.

Applying it

• When applying the heuristic rules, write down for every M deleted or kept exactly which rule justified it and what the key judgment call was, rather than deciding by intuition.
• Present three versions of the estimate side by side — no M inserted at all, M inserted before every operation, and M inserted selectively per the rules — to make the magnitude of the rules' own effect on the result visible at a glance.
• Use skilled users' think-aloud or recorded operation video to locate their real pauses, and compare the rule-based result against what was actually observed, rather than assuming the rules necessarily predict correctly.
• How to check: whenever the interface changes — especially when autocomplete, templates, or hints are added — re-examine whether the M rules still apply, since such changes shift steps that once required the user's own judgment over to the system, and the rules' conclusions should be updated accordingly rather than reusing an old result from before the change.

Related

• Same group: B4.12.1 Tasks are split into standard units such as keystroke, point, home, draw, mental preparation, and system response · B4.12.3 System response time counts toward the total only when it blocks the user · B4.12.4 Absolute constant values are questionable, but the differences between alternatives are relatively reliable · B4.12.5 Touch, voice, and gesture lack agreed-upon constants and must be measured before use
• Nearby: B4.05 Keystroke-Level Model · A6.02 Working Memory Capacity
• Search terms: mental operator placement · KLM heuristics · estimation error · heuristic rule