Response time is increasingly used as a window to look into decision-making process in economic experiments. Yet its interpretation is ambiguous: a fast response may reflect quick comprehension or an impulsive choice. We show that the source of this ambiguity is conceptual and can be resolved by a simple design feature: the separation of instruction and decision screens. Using data from approximately 1,100 students in a lab-in-the-field experiment linked to standardized test scores, we decompose total response time into comprehension time and deliberation time. We show that each component is informative in its own right. Comprehension time is correlated with cognitive ability, providing a cost-free ability proxy in any experiment with split screens. Deliberation time is independent of ability but responds to the structure of the choice: heuristic choices are faster and decisions near the indifference point take longer. The decomposition recovers behavioral patterns that total response time obscures.