The mechanism
Evidence for competing responses accumulates over time until it reaches a decision boundary. A lower boundary produces quicker responses but more errors; a higher boundary produces slower, more cautious responses. Stimulus quality changes how quickly useful evidence accumulates, while instructions and rewards change how much evidence the participant demands. Similar average times can therefore arise from different underlying strategies.
What to measure
Display median correct response time with accuracy and the full count of premature, incorrect, and missed responses. Compare the response-time distributions for correct and error trials rather than keeping only a personal best. If a combined score is needed, define it in advance and show its components so users can see whether a gain came from speed, accuracy, or both. A useful summary also marks difficulty and the instruction given, because speed-focused and accuracy-focused blocks are not interchangeable baselines.
Useful practice conditions
- Set an accuracy floor before awarding speed gains, and explain it before the round begins.
- Alternate neutral, speed-focused, and accuracy-focused blocks to reveal strategy changes.
- Keep difficulty adaptive enough to avoid both ceiling accuracy and random guessing, but change only one difficulty dimension at a time.
Limits of the conclusion
A lower response time is an improvement only relative to a defined accuracy requirement and task difficulty. A combined score is a design choice, not a natural unit of cognitive ability. Gains may remain specific to the trained stimuli, response mapping, and reward structure, so they should not be generalized beyond the measured setting without separate evidence.