Is continuous report more efficient than 2AFC?

[Niko Busch]

Hi everyone,

I am making my first steps with the MemToolbox. This is an awesome tool that is going to make my life a lot easier - thank you guys for making the code available to the community!

I am also in the process of setting up a new experiment, and I am not sure whether it would be better to use a 2AFC procedure or a continuous report procedure. I understand that MemToolbox can estimate models for both types of tasks. Specifically, I wonder: is either task expected to yield more accurate estimates than the other? And does a continuous report procedure require fewer trials to arrive at a reliable model estimate?

Any help regarding these questions would be much appreciated!
Niko Busch

[Jordan Suchow]

Hi Niko,

In general this will depend on the method that you use to select stimulus intensities for the 2AFC task. If, for example, you decide to measure the full psychometric function by testing a wide range of offsets (say, 0 to 100 deg), then you will likely require several hundred trials for accurate measurements, whereas you could make do with less using the continuous-report paradigm. However, there are fancier stimulus placement strategies (e.g., adaptive staircases procedures such as QUEST or FAST) that may improve the efficiency of the testing procedure quite a bit, allowing you to place each trial at the most informative intensity and thereby avoiding less informative regions of the curve (say, anything past 40 deg with set size 1 or 2).

If you'd like a quantitative answer to your question, take a look at MemTests/TestSamplingAndFitting, which simulates data from a model using parameters of your choosing and then attempts to recover those parameters using MemFit.

Hope this helps,

Jordan

-- 

Jordan Suchow

http://suchow.io

--
You received this message because you are subscribed to the Google Groups "MemToolbox users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to memtoolbox-use...@googlegroups.com.
To post to this group, send email to memtoolb...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/memtoolbox-users/bdcd75de-cdf8-40c2-88fa-9e7892f0514f%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.