Optimal Response Initiation: Why Recent Experience Matters
Matt Jones,* Michael C. Mozer,* Sachiko Kinoshita
* University

of Colorado Macquarie University

#T74

Diffusion Models of Decision Making
Evidence
2 1 0 -1 0 50 100

Optimal Inference Depends on Drift Rate
^ µ = .02 Posterior Probability
1 0.5

Drift rate µ
(correct resp.)

^ µ = .04

Drift rate 0
(incorrect resp.)

0 0

50

100

0

50

100

Time

Estimating µ From Past Experience

Sequential Effects of Task Difficulty Response Time

...

an-1,bn-1

an,bn ~ N(an,bn)

µn-1 Rn-1 En-1 Rn

µn En

Trial Sequence (Easy, Hard)