Drift Diffusion Model
Also known as: DDM, Brownian Motion Model, Sequential Sampling Model
The Drift Diffusion Model (DDM) is a mathematical framework for understanding rapid binary decision-making by modeling the accumulation of evidence over time as a random walk with drift. Developed by Roger Ratcliff in the 1970s, it predicts both choice probabilities and response time distributions, providing insight into the cognitive processes underlying decisions in perceptual discrimination, recognition memory, and choice tasks.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use the Drift Diffusion Model when analyzing binary choices with both accuracy and response times available, especially in perceptual discrimination, lexical decision, recognition memory, or any speeded two-choice task. It is particularly valuable when examining individual differences in processing speed or cautiousness, when comparing drift rates across experimental conditions, or when fitting model parameters reveals latent cognitive processes.
Strengths & limitations
- Unified framework capturing both accuracy and response time, providing richer information than accuracy alone
- Interpretable parameters (drift rate, boundary separation, non-decision time) map onto cognitive constructs (evidence quality, decision cautiousness, motor delays)
- Applies broadly across decision domains: perception, memory, categorization, and value-based choice
- Enables formal hypothesis testing through model comparison and parameter recovery
- Model complexity makes fitting computationally intensive; requires specialized software and statistical expertise
- Assumes constant drift rate and single-stage accumulation; violations (time-varying evidence, multiple stages) require extended models
- Requires data from many trials (typically 50+ per condition) for stable parameter estimates
- Limited theoretical guidance for choosing between variants (standard, full diffusion, hierarchical versions)
Frequently asked
What does drift rate tell me about the decision process?
Drift rate (v) measures how quickly, on average, evidence accumulates toward the correct choice. High drift rates indicate strong or clear evidence, fast information processing, or ease of discrimination. Low drift rates indicate weak evidence, slow processing, or difficulty. By comparing drift rates across conditions (e.g., high vs. low stimulus contrast), you reveal what factors affect evidence quality.
How do boundary separation and cautiousness relate?
Boundary separation (a) represents the amount of evidence needed to decide. Larger boundaries mean the decision-maker requires more evidence before responding, making them more cautious and accurate but slower. Smaller boundaries mean faster but less accurate decisions. Non-decision time (Ter) captures perception and motor delays, independent of the decision process.
Why do I need response time data, not just accuracy?
Response times provide information about the diffusion process that accuracy alone cannot reveal. An observer with 75% accuracy could have high drift rate but require extensive evidence (large boundary), or low drift rate with minimal evidence needed. Response time distributions disambiguate these scenarios.
Can I fit the Drift Diffusion Model without specialized software?
Formal fitting requires numerical methods (likelihood estimation or Bayesian approaches). Free software includes HDDM (Python, hierarchical Bayesian), fast-dm, and EZ-diffusion. Existing software packages in R and Python handle most applications; fitting from scratch is not recommended without statistical expertise.
Sources
- Ratcliff, R. (1978). A theory of memory retrieval. Psychological Review, 85(2), 59-108. DOI: 10.1037/0033-295X.85.2.59 ↗
- Ratcliff, R., & McKoon, G. (2008). The diffusion model: A universal model for rapid decision. Psychological Review, 115(2), 357-380. link ↗
- Wagenmakers, E.-J., van der Maas, H. L. J., & Grasman, R. P. P. P. (2007). An EZ-diffusion model for response time and accuracy. Psychonomic Bulletin & Review, 14(1), 3-22. DOI: 10.3758/BF03194023 ↗
How to cite this page
ScholarGate. (2026, June 3). Drift Diffusion Model. ScholarGate. https://scholargate.app/en/psychology/drift-diffusion-model
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Signal Detection TheoryPsychology↔ compare