Response deadline methods are experimental techniques that control the speed-accuracy tradeoff by requiring participants to respond within a specified time window. By systematically varying the deadline across conditions, researchers can trace out the time course of information processing — observing accuracy rise from chance at very short deadlines to asymptote at long deadlines. These methods complement free-response paradigms by providing direct experimental control over processing time.
Types of Deadline Methods
Several variants of the deadline methodology have been developed, each with distinct advantages:
Response deadline: Response must occur before a cutoff; late responses discarded
Speed-accuracy manipulation: Block-level instructions emphasizing speed vs. accuracy
Response signal method: A signal (typically an auditory tone) is presented at a variable lag after the stimulus, and participants are trained to respond within a narrow window (~200 ms) after the signal. This is the most precise method, yielding a fine-grained mapping of the SAT function. Pioneered by Reed (1973) and refined by Wickelgren and colleagues, it produces clean time-course data.
Hard deadline: Participants are given a fixed time window (e.g., 300, 500, 800, 1500 ms) and must respond before it expires. Responses after the deadline are discarded or penalized. This method is simpler to implement but produces less precise time-course data because response times vary within the window.
Methodological Considerations
Deadline methods face several challenges. Participants must be extensively trained to respond within the required time window; early in training, participants often fail to comply with very short deadlines or shift to a guessing strategy. The response signal method requires careful calibration of the acceptable response window — too narrow and participants cannot comply, too wide and processing time is poorly controlled.
A critical analytical choice is whether to analyze accuracy conditional on the actual response time (yielding a CAF-like function) or unconditional (simply computing accuracy at each deadline lag). Conditional analysis can introduce artifacts because it conditions on a variable (RT) that is itself affected by accuracy, creating a selection bias. Unconditional analysis avoids this bias but provides less temporal precision. McElree and Dosher (1989) advocated for the unconditional approach when mapping SAT functions.
Applications and Findings
Response deadline methods have been applied across many domains. In recognition memory, they have revealed that familiarity-based information becomes available before recollection-based information (Dosher, 1984), with familiarity reaching asymptote by ~750 ms and recollection contributing accuracy gains only after ~1000 ms. In sentence processing, McElree (2000) used the response signal method to show that syntactic violations are detected rapidly but semantic anomalies require additional processing time.
In visual search, deadline methods have helped distinguish parallel from serial processing: parallel search shows rapid accuracy growth across all set sizes, while serial search shows progressively delayed accuracy growth as set size increases, consistent with items being searched sequentially. These applications demonstrate the power of deadline methods for resolving fundamental questions about the architecture and time course of cognitive processes.