What Happens to Judgment and Accuracy When Time Runs Out?
Decision-making under time pressure is the selection of an action, belief, or option when an actual or perceived deadline restricts how long the decision maker can acquire, evaluate, or integrate information. The constraint can alter the amount of evidence processed, the rule used to compare alternatives, the willingness to accept uncertainty, and the point at which a decision terminates. A deadline sometimes reduces accuracy, yet it can also motivate efficient information search or preserve an opportunity that would disappear after delay. The central scientific finding is that the effects depend on the task and on what counts as a successful decision. Payne et al. (1988); Heitz (2014).
A patient-facing team deciding whether to escalate an urgent condition, an operator choosing a response to an alarm, an analyst finalizing a forecast, and a shopper deciding before a sale expires all face constrained time. Their optimal policies differ because delays, errors, and missed opportunities have different consequences. Rapid action is often necessary in changing environments; more deliberation can be worthwhile when additional evidence is informative and an error is costly. Understanding time pressure therefore requires joint measurement of response time, outcome quality, missed deadlines, and the cost structure of the problem.
This chapter examines psychological mechanisms, classical experiments, normative models, modern research, risk preference, expertise, reproducibility, and ways to improve decisions when time is scarce. It separates empirical evidence about specific tasks from broad claims about intuition or rationality. An observed change in choice under a deadline is a behavioral finding; identifying the responsible cognitive mechanism requires additional evidence.
Defining Time Pressure Precisely
Objective deadlines, perceived urgency, and uncertainty
An objective deadline specifies the latest time at which a response can be accepted. A relative constraint means that the person has less time than the task seems to require, even when no hard cutoff exists. Subjective urgency is the sense of needing to hurry. These conditions can overlap but are distinct experimental manipulations. A countdown timer can create a predictable limit; an unpredictable interruption can produce urgency without a known terminal moment; a difficult task can feel rushed despite a relatively long window. Studies must describe the manipulation rather than use the label time pressure as though it defined one uniform intervention.
A hard deadline differs from instructions merely to answer quickly. A missed hard deadline can invalidate a response, while speed instructions may change a performance bonus without excluding late answers. Response-signal paradigms control how long evidence is accessible before a required response. Field studies often add shifting evidence, teamwork, delayed feedback, and consequences that continue after the choice. An experiment with an 800-millisecond deadline and a professional judgment lasting several minutes therefore addresses related but nonidentical phenomena. Karşılar et al. (2014).
Accuracy, rationality, and the chosen benchmark
Accuracy in perceptual discrimination means agreement with a known stimulus category; accuracy in forecasting means closeness between reported probabilities and later outcomes; and decision quality concerns the value of an action under a defined objective. A person can make a well-calibrated estimate and still suffer an unfavorable outcome, or choose a risky option despite correctly estimating the probabilities. The normative standard might be probability theory, a proper scoring rule, expected monetary value, expected utility, or a domain-specific safety requirement. The standard must be specified before labeling a choice an error.
An example reveals why the distinction matters. One responder gets 95% of 40 classifications correct in ten minutes, yielding an expected 38 correct; another gets 80% of 60 correct, yielding an expected 48. Under a goal of maximizing total correct classifications with no additional error penalty, the faster policy yields the larger expected total. Under a goal of maximizing percentage correct, the slower policy wins. Under severe costs for errors the preferred policy may change again. These figures are illustrative arithmetic and are not research effect-size estimates. Formal optimality arguments depend on the performance objective and associated assumptions. Bogacz et al. (2006).
Different concepts must remain separate
Time pressure is a condition of decision-making, not a cognitive bias, personality trait, logical fallacy, or clinical diagnosis. The speed–accuracy tradeoff is a relation between response policies, reaction times, and errors that is most clearly measured in repeated discrimination tasks. Cognitive processing speed is an individual ability or task-dependent performance measure. Cognitive load concerns processing demands; stress includes wider affective and physiological responses. These conditions may interact, but the effects of one cannot automatically be attributed to another. Time pressure also differs from decision fatigue, which concerns repeated decisions or changing engagement over time.
The History of Time-Constrained Decision Research
Evidence accumulation and sequential statistical analysis
Reaction-time research and sequential decision theory developed the idea that people can continue acquiring evidence until a stopping condition is reached. In simplified two-choice tasks, waiting longer may increase the reliability of noisy evidence, while delay has an opportunity cost. Models of sequential sampling formalize that tension through response boundaries and a stochastic accumulation process. Bogacz and colleagues analyzed conditions under which particular thresholds optimize a specified rate of reward and compared alternative decision processes. Such mathematical results depend on their assumptions; they do not imply that one threshold is optimal for all human judgments. Bogacz et al. (2006).
Formal accounts remain useful precisely because they force researchers to distinguish the quality of incoming evidence from the willingness to respond on limited evidence. A rapid correct answer can result from clear evidence; an equally rapid incorrect answer can result from noisy evidence or premature commitment. Without measures of the information stream and behavior across trials, these explanations remain confounded.
Adaptive strategy selection
Payne, Bettman, and Johnson examined how people select decision strategies as task demands and time pressure change. In their influential 1988 experiments and simulations, heuristics were evaluated for computational effort and how closely they approximated more elaborate procedures. Participants adapted information-search behavior to properties of the alternatives and the time available, broadly consistent with effort–accuracy tradeoffs. The study also demonstrates why a simpler strategy cannot automatically be classified as an irrational one: no single procedure performed best under every task structure. Payne et al. (1988).
A weighted-additive strategy scores several attributes and lets strength on one attribute compensate for weakness on another. A lexicographic strategy first checks the most important attribute and turns to secondary attributes only when necessary. Elimination-by-aspects removes alternatives that fail selected criteria. These procedures have different informational requirements. Under severe pressure, the choice of procedure can be as consequential as the eventual choice among options.
Dynamic preferences and naturalistic expertise
Decision field theory models preferences as processes that evolve while attention and comparisons fluctuate. It explains several choice and timing phenomena, including preference changes under time constraints, without treating observed preferences as static quantities fixed before deliberation. Busemeyer and Townsend (1993).
Naturalistic decision research examines experts in environments in which information is incomplete and consequences unfold rapidly. Klein and colleagues studied experienced fireground commanders and reported evidence of simultaneous comparative evaluation of multiple options at fewer than 12% of the documented decision points. Commanders commonly identified a plausible response through recognition of a situation and assessed whether it would work. This field observation is informative about expertise in the sampled incidents; it is not an estimate of the same pattern in all professions. Klein et al. (1986).
How Deadlines Alter Cognition
Decision thresholds and the diffusion model
The diffusion decision model explains decisions between two alternatives as the accumulation of noisy evidence toward one of two boundaries. Drift rate reflects the average strength and direction of the evidence. Boundary separation reflects the amount required to commit. Starting position can capture prior preference, while nondecision time represents components such as encoding and motor execution. In common task conditions, a lower boundary leads to faster responses and more errors while a higher boundary permits more evidence accumulation and higher accuracy. Ratcliff and McKoon (2008).
A central limitation is that behavior cannot be identified from one mean reaction time or one error rate. Shorter latencies can arise from improved evidence quality, lower caution, motor preparation, or shifts in attention. Physiological experiments by Heitz and Schall showed that speed–accuracy adjustments can involve multiple neural changes, including sensory processing and response preparation. Consequently, claims that time pressure simply lowers a decision threshold must be tested against alternative accounts. Heitz and Schall (2012).
Information search and selective attention
When a decision has many attributes, time is spent determining what to inspect as well as evaluating what has already been seen. Under pressure, people can reduce the breadth of search, prioritize decisive cues, stop at the first satisfactory alternative, or reuse a familiar procedure. The resulting choice can remain effective when omitted attributes contribute little information. In contrast, an omitted safety-critical attribute may transform an efficient shortcut into an expensive mistake. This dependence on cue validity and task structure lies at the center of adaptive strategy research. Payne et al. (1988).
An experiment may assess search by recording the sequence of information requests, attribute fixation, mouse movement, eye gaze, or the number of alternatives inspected before choice. Process data matters because the same final answer can be produced by careful selective search or arbitrary early stopping. A deadline manipulation should be tested for its effects on both information acquisition and the final decision.
Strategy carryover and familiarity
Ordóñez and Benson studied judgments of risky gambles made under time limits. In particular task sequences, participants carried a strategy from an earlier pricing judgment into a subsequent attractiveness rating, producing systematic changes in expressed preferences. These effects depended on prior task experience and participant characteristics. The results support a conditional account of strategy reuse, not a universal shift toward impulsiveness. Ordóñez and Benson (1997).
The practical lesson is that familiarity can improve rapid decisions only when learned cues and procedures remain appropriate. A well-practiced emergency protocol may save time; a familiar response imported into a newly configured system may produce a confident mistake. Research must examine whether the learned mapping remains valid.
Emotion and motivation
Time constraints can create anxiety, arousal, and intensified effort, each of which may change attention and strategy. Maule, Hockey, and Bdzola manipulated decision deadlines in risky scenarios that differed in outcome valence and the effort associated with a safer action. Time-pressured participants reported greater anxiety and energy, while changes in risk-taking and strategy varied with the task structure. This pattern argues for measuring both affect and behavior rather than attributing every deadline effect to an assumed emotional response. Maule et al. (2000).
Pressure can lead decision makers to protect their primary goal while dropping secondary verification tasks. Such prioritization is sometimes an appropriate allocation of limited effort. It becomes dangerous when a supposedly secondary check detects an unlikely but catastrophic failure. The value of the check depends on the probability and cost of that failure, and the alternative actions available after an error is discovered.
Normative Decisions and Deadlines
Correct answers per trial and per minute
Maximizing accuracy on each trial differs from maximizing correct answers in a limited session. Formal models can identify optimal stopping boundaries for particular objective functions, information qualities, and intertrial delays. When late decisions receive no reward, waiting too long may erase the value of additional evidence. When errors are dangerous or irreversible, greater caution can be optimal. There is no single ideal response time independent of the payoff structure. Bogacz et al. (2006).
The normative model must also distinguish the value of a correct action from the value of knowing the truth. A correct conclusion reached after a window for action has closed may possess informational value but little immediate practical utility. Conversely, a quick intervention may have high expected value even when some uncertainty remains.
Do people lower thresholds as deadlines approach?
An idealized decision maker facing a fixed cutoff may gradually lower the evidence requirement near the endpoint to avoid a missed response. Such a collapsing-bound strategy has theoretical appeal, but its empirical implementation is a distinct question. Karşılar, Simen, Papadakis, and Balcı studied random-dot motion decisions with free-response, 800-millisecond, and 1,000-millisecond deadline conditions. Their observations did not match several predictions about optimal within-trial threshold collapse and could be reproduced by a model with fixed within-trial parameters. Karşılar et al. (2014).
This result constrains general claims about deadline anticipation. The capacity to adjust a response policy between trials does not establish that people continuously readjust it within every very short trial. The distinction also matters when timing is uncertain and the probability of a missed cutoff changes over time.
Reaction time is not a direct measure of reasoning mode
A fast response does not demonstrate that the person relied on intuition, and a slow response does not demonstrate analytic correction. Difficult alternatives are often both slower to discriminate and more likely to produce error even under the same processing mechanism. Krajbich and colleagues criticized reverse inference from latency to separate intuitive and deliberate mental systems, demonstrating why differences in discriminability must be considered. Krajbich et al. (2015).
Dual-process theories can motivate testable predictions, but the label fast thinking cannot replace a process model. Studies should consider choice difficulty, confidence, information search, learning, and response-time distributions before drawing claims about the architecture of reasoning.
Does Time Pressure Increase Risk Taking?
A major meta-analysis finds both directions
Belli, Carrillat, Zlatevska, and Cowley synthesized 213 effect sizes across 83 papers and 65,574 unique respondents. In the included findings, 38.5% indicated that time pressure increased risk preference whereas 61.5% indicated the opposite direction. These percentages describe the direction of reported effects, not the shares of individual people who became more risk seeking or more risk averse. The meta-analysis identified moderators involving the nature of the constraint, negative outcome salience, severity, and vulnerability. It provides strong evidence against the claim that time pressure invariably increases risk taking. Belli et al. (2024).
Whether a deadline is absolute or relative, objective or ambiguous, can alter the measured response. So can the framing of gains and losses and whether risk was learned through description or experience. Researchers also distinguish hypothetical attitudes toward risk from actual choices with consequences. Results aggregated from heterogeneous designs cannot be transferred automatically to professional emergency decisions.
Risk preference differs from probability error
Choosing a riskier option can reflect a preference about outcomes rather than a distorted probability judgment. To identify a bias, the analyst must establish an appropriate benchmark, such as dominance, expected value, or a utility model consistent with the decision maker's goals. Two people can agree on the likelihood of a loss yet choose differently because the consequences matter differently to them.
A deadline can change attention to reward magnitude, likelihood, and downside severity, potentially shifting expressed preference even when basic beliefs remain similar. Controlled studies of gamble valuation demonstrate conditional changes in how information is combined. Ordóñez and Benson (1997). An inference that all quick risks are impulsive errors neglects these distinctions.
Expertise and Time Pressure in Real Environments
Recognition-based decisions
Experienced responders may quickly identify characteristic patterns and a plausible course of action. Naturalistic field research supports the importance of recognizing familiar situations without exhaustively comparing every theoretical alternative. This is one reason highly practiced experts can perform well under short decision windows, particularly when cues are stable and feedback has made the relevant patterns learnable. Klein et al. (1986).
Expertise must be understood within its field. High performance in familiar situations is not proof of superior judgment across unrelated tasks. The most difficult cases are those in which a rare event resembles a common one, a previously diagnostic cue becomes unreliable, or the environment has changed without corresponding updates to training. Effective systems preserve mechanisms for detecting anomalies and escalating uncertainty.
Team decisions and organizational design
High-stakes rapid decisions often occur in groups rather than isolated individual minds. Communication requirements, distribution of expertise, role clarity, and the availability of reliable records can matter as much as one person's response threshold. A deadline may make coordination more costly, yet carefully established protocols can reduce redundant work and prevent critical omissions. Organizational interventions should be evaluated using domain-specific outcome data instead of assuming that laboratory findings about single-person classification are sufficient.
A high-consequence decision is also embedded in recovery opportunities. Some choices are reversible and permit later correction; others are effectively permanent. Rational urgency therefore includes knowing whether the next action is a commitment, a temporary safeguard, a request for more information, or a means of preserving options.
Measuring Time-Pressure Effects Scientifically
Experimental design
A rigorous study defines the deadline, task difficulty, reward contingencies, number of alternatives, information availability, and participants' familiarity with the procedure. Randomization or counterbalancing helps distinguish pressure from task order and practice effects. Researchers should verify that the time-pressure manipulation actually changed behavior or perceived urgency. Otherwise, a statistically insignificant result may simply reflect a weak manipulation.
Deadlines create a special missing-data problem. If late responses are excluded from accuracy analysis, the remaining completed answers may appear more accurate than the full policy's performance. Correct responses, incorrect responses, omissions, deadline violations, and timeouts are separate outcomes. Their frequencies and confidence intervals should be reported, ideally alongside trial-level reaction times.
Modeling full distributions
Averaging reaction times obscures differences between fast errors, slow errors, and changing decision policies. Diffusion models use choices and response-time distributions to estimate latent components, but their interpretation depends on model fit, parameter identifiability, and alternative specifications. A parameter labeled caution is a model estimate rather than a direct clinical or neurological observation. Ratcliff and McKoon (2008); Heitz (2014).
For complex choices, process tracing can measure the number and order of inspected attributes. Combining this with accuracy and response timing helps determine whether a deadline changed information quality, search strategy, or stopping policy. Retrospective reports may complement these records but can omit important steps or provide post hoc explanations.
Reliability, replication, and generalization
Evidence for controlled speed–accuracy relations is established across many discrimination paradigms. Generalizing to dynamic decisions with multiple options is a separate research challenge. Laboratory clocks may not reproduce professional accountability, injury risk, teamwork, or feedback delays. Decision researchers should replicate effects under comparable task definitions and test generalization to novel scenarios. Heitz (2014).
A model that explains one response-time distribution may have competitors that fit almost equally well. Stronger evidence comes from predicting new conditions, manipulating the proposed mechanism, and comparing models on held-out data. Meta-analytic averages must be interpreted with the heterogeneity of tasks and populations in view. Findings from professional incidents can be ecologically rich while requiring care about sampling and reconstruction from interviews.
Improving Judgment When the Clock Is Running
Identify which errors and delays matter
The first step is to define the objective: maximize expected benefit, prevent rare severe harm, preserve an opportunity, or reach a sufficiently accurate judgment before a deadline. The optimal allocation of time follows from the cost of delay and the value of another observation. A universal instruction to slow down can be harmful when action has to precede an irreversible change. An equally universal instruction to decide instantly can remove essential safety checks. Bogacz et al. (2006).
Prioritize diagnostic information
Decision makers can prioritize the evidence most likely to alter the preferred action. Valid exclusion rules, ranked attributes, and clearly defined stopping criteria reduce effort when the choice structure permits them. Training should teach not only a heuristic's normal application but also the conditions under which its cues fail. The effort–accuracy framework supports such strategy selection as a design principle; specific improvement claims require trials in the target task. Payne et al. (1988).
For example, under an urgent operational deadline, a short protocol can ask which hazard is irrecoverable, which facts most strongly distinguish options, whether a temporary protective action can buy time, and what observation would change the recommendation. The answers should be domain-tested and practiced before an emergency. A new complex checklist introduced during a crisis may add workload rather than reduce it.
Prepare rules before urgency
Precommitted thresholds, standard response procedures, warning triggers, and escalation criteria can spare people the need to reconstruct decisions while rushed. The value lies in evidence-based selection of those rules and preservation of paths for exceptions. A protocol that simplifies ordinary cases but misses low-frequency catastrophic conditions may be unsuitable for a safety-critical system. Checklists should be audited against both common and rare consequential failures.
Train for transfer rather than rehearsal alone
Practice can improve familiarity, speed, and cue recognition. The critical outcome is performance on relevant unfamiliar cases after the training context changes. Repetition of one scenario can make a participant fast on that scenario without establishing robust decision competence. Strong evaluations include retention, transfer to novel problems, errors under surprise changes, and field outcomes. Claims that one generic debiasing course solves hurried decision errors across professions exceed the available evidence.
Calibrate human–technology decision support
Digital tools and AI systems can shorten search and computation time, expose comparisons, and document assumptions. They can also deliver fluent but wrong summaries, omit rare hazards, or encourage overreliance when users have little opportunity to verify an answer. A sound experiment measures the performance of the human–technology team, including detection of erroneous recommendations, false alarms, time saved, and the cost of failures. A fast output should communicate uncertainty and traceable evidence in proportion to the consequences of error.
Human–AI cognition concerns the quality of joint judgment; the subjective-experience status of an AI system is a separate philosophical and scientific question. The present evidence about time pressure supports task-specific study of reliance and verification, not assumptions about consciousness.
Examples and Applications
Emergency decisions
Emergency workers operate with incomplete reports, variable expertise, changing hazards, and costly delays. Recognition-based expertise can support rapid first actions, but unusual signals require escalation. Measured quality must include safety outcomes, correct prioritization, delayed actions, omitted information, and the opportunity for recovery. Findings from one fireground sample inform hypotheses but cannot supply universal performance rates for every emergency occupation. Klein et al. (1986).
Consumer and financial choices
Sales countdowns and imminent market events impose constraints on comparing outcomes. The meta-analytic evidence indicates that changes in risk preference depend on how the constraint is experienced and how losses are framed. Consumers and institutions can evaluate whether a deadline reflects a genuine opportunity cost or creates urgency disconnected from the underlying choice quality. Belli et al. (2024).
Forecasting and evidence evaluation
An analyst may have to deliver a probability estimate at a fixed moment despite incomplete evidence. The normative task involves calibration and decision relevance, not certainty about the final outcome. An effective procedure records the evidence, the estimated probability, the principal missing facts, and conditions that would change the estimate. Later evaluation should use multiple scored forecasts rather than infer ability from a single correct prediction.
Interface and AI-assisted decisions
A time-sensitive interface can improve performance when it reduces unnecessary navigation and makes decisive information legible. It can undermine performance when it hides uncertainty, highlights superficial cues, or pressures users into irreversible confirmation. Systems should be tested on representative scenarios and accessible across relevant user groups. The fact that a recommendation appears earlier is evidence of speed, not automatically of better judgment.
What Remains Uncertain?
The broader evidence supports reliable speed–accuracy relations in controlled tasks, observed adaptation of decision strategies, and heterogeneity of risk preference under pressure. Open questions concern how often decision bounds truly change within a short deadline, which kinds of domain-specific expertise generalize to novel situations, and how structured aids improve consequential real-world outcomes. Karşılar et al. (2014); Payne et al. (1988); Belli et al. (2024).
Potential moderators include stakes, deadline predictability, familiarity, attention, literacy, feedback quality, language, social pressure, and institutional rules. Cultural generalization requires sampling across actual populations and examining whether the same tasks and incentives mean the same thing in each setting. The absence of a universal effect direction is itself scientifically important and should guide both replication and policy design.
Frequently Asked Questions
Does time pressure always reduce accuracy?
It often reduces available evidence or encourages earlier stopping, but its outcome depends on the value of speed, the diagnostic information already available, and task difficulty. Familiar and clear decisions may remain accurate under short deadlines. Heitz (2014).
Does a quick decision prove intuition rather than reasoning?
No. Response time depends on discriminability and processing conditions; quick choices can be well supported, while slow choices can remain uncertain. Reaction time alone cannot identify a psychological system. Krajbich et al. (2015).
Do people become more risk taking under pressure?
Some do in particular designs, while other conditions show increased caution. A meta-analysis of 83 papers found meaningful effects in both directions with identifiable moderators. Belli et al. (2024).
What is an adaptive decision strategy?
It is a procedure selected for the information structure, available time, and objectives of a task. A short rule can preserve much of the quality of a longer calculation when its cues are valid. Payne et al. (1988).
Is a deadline the same as stress?
A deadline is a constraint on available time, whereas stress encompasses broader responses to demands. The two may interact, but they should be measured independently. Maule et al. (2000).
Why can experts decide so quickly?
Relevant experience can support rapid recognition of informative patterns and feasible actions. Expertise is contingent on the environment matching learned cues and can fail with unusual or changing conditions. Klein et al. (1986).
What is the best decision under time pressure?
Best refers to a specified criterion: correctness, expected utility, reward per unit time, avoidance of severe harm, or timely completion. Different criteria can imply different optimal actions and stopping rules. Bogacz et al. (2006).
Can training remove deadline-related errors?
Training can improve practiced actions and sometimes transferable skills, but broad generalization must be demonstrated with novel tasks and real outcomes. A training gain on one task does not establish universal debiasing or elimination of speed–accuracy constraints.
Scientific Synthesis
Decision-making under time pressure is the regulation of evidence search, evaluation, commitment, and action within a finite opportunity window. Its most reliable lesson is that optimality must be defined with both error costs and delay costs in view. Experimental models describe some mechanisms precisely; adaptive strategy research and naturalistic studies show how task structure and experience shape broader choices. Scientific accuracy requires measuring the whole decision process and resisting universal claims about haste, risk, intuition, and competence.
Related Articles
Cognitive Debiasing Strategies: Which Interventions Work? — Research on training, feedback, and corrective decision procedures.
Decision-Making Under Risk and Uncertainty: Models and Examples — Expected-value, utility, and descriptive models of choice.
Risk Perception: How People Judge Hazards and Probabilities — How people assess likelihood and severity of hazards.
Speed-Accuracy Tradeoff: Why Faster Responses Can Mean More Errors — The focused response-time and evidence-accumulation literature.
Naturalistic Decision-Making: How Experts Choose Under Pressure — Studies expert recognition and situational assessment when rapid decisions have real consequences.
References
Belli, A., Carrillat, F. A., Zlatevska, N., and Cowley, E. (2024). How does time pressure influence risk preferences? Answers from a meta-analysis. Journal of Consumer Research, 50(6), 1172–1197. https://doi.org/10.1093/jcr/ucad053
Bogacz, R., Brown, E., Moehlis, J., Holmes, P., and Cohen, J. D. (2006). The physics of optimal decision making: A formal analysis of models of performance in two-alternative forced-choice tasks. Psychological Review, 113(4), 700–765. https://doi.org/10.1037/0033-295X.113.4.700
Busemeyer, J. R., and Townsend, J. T. (1993). Decision field theory: A dynamic-cognitive approach to decision making in an uncertain environment. Psychological Review, 100(3), 432–459. https://doi.org/10.1037/0033-295X.100.3.432
Heitz, R. P. (2014). The speed-accuracy tradeoff: History, physiology, methodology, and behavior. Frontiers in Neuroscience, 8, 150. https://doi.org/10.3389/fnins.2014.00150
Heitz, R. P., and Schall, J. D. (2012). Neural mechanisms of speed-accuracy tradeoff. Neuron, 76(3), 616–628. https://doi.org/10.1016/j.neuron.2012.08.030
Karşılar, H., Simen, P., Papadakis, S., and Balcı, F. (2014). Speed accuracy trade-off under response deadlines. Frontiers in Neuroscience, 8, 248. https://doi.org/10.3389/fnins.2014.00248
Klein, G., Calderwood, R., and Clinton-Cirocco, A. (1986). Rapid decision making on the fire ground. Proceedings of the Human Factors Society Annual Meeting, 30(6), 576–580. https://doi.org/10.1177/154193128603000616
Krajbich, I., Bartling, B., Hare, T., and Fehr, E. (2015). Rethinking fast and slow based on a critique of reaction-time reverse inference. Nature Communications, 6, 7455. https://doi.org/10.1038/ncomms8455
Maule, A. J., Hockey, G. R. J., and Bdzola, L. (2000). Effects of time-pressure on decision-making under uncertainty: Changes in affective state and information processing strategy. Acta Psychologica, 104(3), 283–301. https://doi.org/10.1016/S0001-6918(00)00033-0
Ordóñez, L., and Benson, L., III. (1997). Decisions under time pressure: How time constraint affects risky decision making. Organizational Behavior and Human Decision Processes, 71(2), 121–140. https://doi.org/10.1006/obhd.1997.2717
Payne, J. W., Bettman, J. R., and Johnson, E. J. (1988). Adaptive strategy selection in decision making. Journal of Experimental Psychology: Learning, Memory, and Cognition, 14(3), 534–552. https://doi.org/10.1037/0278-7393.14.3.534
Ratcliff, R., and McKoon, G. (2008). The diffusion decision model: Theory and data for two-choice decision tasks. Neural Computation, 20(4), 873–922. https://doi.org/10.1162/neco.2008.12-06-420