search

UMPsychMethodsCore / MethodsCore

All of the projects that the methods core develops, combined into one repository!
7 stars 0 forks source link

Paradigm Optimization #35

Closed dankessler closed 4 years ago

dankessler commented 12 years ago

Opening up this thread for @sripada

Discussion should follow on what sort of paradigm optimization we want to support...

rcwelsh commented 12 years ago

I) Optimizations:

A) optimization for contrast efficiency ( I can get the reference for this) B) optimization for T-tests? C) optimization for deconvolution

II) Design types

1) Definitely block design, with A,B,...N types of blocks with potential varying block length and varying baseline length 2) event related design, with N-trial types. Be able to specify the ITI for each trial type such that each trial type has equal integrates ITI. Also, be able to make rules on number of trials in a row.

dankessler commented 12 years ago

I wanted to tag @heffjos on this since he volunteered to work on this.

dankessler commented 12 years ago

As I understand it, this will generally be a two step process.

  1. Generate random design within constraints
  2. Evaluate "fitness" of design. This may be calculated as...
    • Contrast efficiency
    • Particular t-tests of interest
    • Ease of deconvolution

The first part is either really easy or rather challenging depending on the constraints and may vary a bit from study to study. However, we (and by we I mean @heffjos) should be able to write functions (or borrow from elsewhere) that calculate values for each of those three fitness parameters.

heffjos commented 12 years ago

@rcwelsh @sripada @dankessler

I have found a tool for automatically scheduling events for rapid-fire presentation event related fMRI experiments. It is called optseq and is found at http://surfer.nmr.mgh.harvard.edu/optseq/ . It seems pretty flexible.

Some points about it are: