The Problem of Uncertainty
Every day, we make thousands of decisions based on noisy, incomplete information. Should you cross the street? Is that blurry figure a friend or a stranger? In the 1970s, psychologist Roger Ratcliff formalized a model to explain how we make these choices. This is the Drift Diffusion Model (DDM).
How the Model Works
The DDM assumes that decision-making is a process of evidence accumulation. Imagine a particle starting at a neutral point (zero). As you look at the moving dots above, your brain extracts small samples of information.
The Mathematical Foundation
The position of the "evidence" at any time \(t\) is described by a stochastic differential equation:
- \(v\) (Drift Rate): The average speed of accumulation. It represents the quality of the signal.
- \(\sigma\) (Diffusion): The noise or randomness added at each step.
- \(dW\): A Wiener process (white noise).
The Three Key Parameters
1. Coherence (The Drift Rate)
In the "Random Dot Kinematogram" stimulus above, coherence is the percentage of dots moving in the same direction. In the brain, this is represented by neurons in the MT (Middle Temporal) area. High coherence creates a strong "Drift," pushing the evidence quickly toward a boundary.
2. Boundary Separation (The Threshold)
The boundaries (the green and red lines) represent the amount of evidence required to commit to a choice. If you are cautious, your brain sets these boundaries far apart. If you are in a rush, you bring them closer. This leads to the famous Speed-Accuracy Tradeoff:
- Narrow Boundaries: Faster decisions, but higher chance of noise accidentally pushing you into the wrong choice.
- Wide Boundaries: Slower decisions, but more robust against random fluctuations.
3. Noise (The Diffusion)
Noise is the "zig-zag" in the path. It comes from two places: the external world (the random dots) and the internal brain (neural firing variability). Without noise, you would never make a mistake on a stimulus with a positive drift rate. Noise is the reason we sometimes choose "Down" even when the dots are clearly moving "Up."
Neural Integration
Research in primates (notably by Newsome and Shadlen) found that neurons in the Lateral Intraparietal (LIP) area actually behave like the evidence accumulator in this model. Their firing rates increase steadily over time, exactly matching the predicted paths of the DDM, until they reach a certain "threshold" frequency that triggers a motor response.