Jitter Papers

Useful Papers - Jitter

Also check out DesignPapers, ScanningPapers, FmriPhysicsPapers and PhysiologyPapers...

Primary:

Liu (2004), "Efficiency, power and entropy in event-related fMRI with multiple trial types, Part II: design of experiments," NeuroImage 21, 401-413 PDF

Summary: The more intelligible part in a two-part manifesto. Liu expands on his 2001 paper (see DesignPapers) to demonstrate the tradeoff betwen efficiency and power is fundamental for multiple-trial-type experiments. Crucially, he also explores what specific experimental designs bring you closest to the theoretical best power for a given efficiency (or vice versa). Permuted block designs look like great happy mediums.

Bottom line: 2-block designs and m-sequence designs are at the opposite ends of the power/efficiency spectrum, each getting you close to the theoretical limit of how much power/efficiency you can get. Permuted block designs give you excellent tradeoffs between them.

Supplementary:

Dale (1999), "Optimal experimental design for event-related fMRI," Human Brain Mapping 8, 109-114 PDF

Summary: Perhaps the best early defense of variable-ISI studies; Dale attempts to end the debate between long event-related designs and rapid event-related designs by showing the variable-ISI designs tremendously improve on the power of rapid designs. A mathematical argument about statistical efficiency is used to show that random-ISI (i.e., jittered) designs can outperform fixed-ISI designs for the same mean ISI by more than a factor of 10.

Bottom line: Once and for all: if possible, you should definitely be using a variable-ISI rapid design in any event-related study. The math is all here...

Friston et. al (1999), "Stochastic designs in event-related fMRI," NeuroImage 10, 607-619 PDF

Summary: Another look at which designs make the most sense - block, fixed-ISI, or variable-ISI (connected to the "stochastic" design idea). Authors make the point that the best type of design will depend very heavily on what type of effect you're looking for - differential effects between conditions do best with a different design than do evoked responses changing from baseline.

Bottom line: An early distinction between efficiency and power is drawn, as a difference between looking for differential or evoked effects. The idea of constructing variable-ISI designs from "null events" is brought forth.


CategoryPapers