New MLP (with two improvements) (Massimo Grassi )

Subject: New MLP (with two improvements)
From:    Massimo Grassi  <massimo.grassi@xxxxxxxx>
Date:    Tue, 11 Nov 2008 17:24:32 +0100

Dear all, (apologies for this further MLP message). we modified the last version of MLP according to the suggestions we received. In the MLP beta web page (see link below) you can now download for testing the new MLP. The modifications of this new MLP are described there and at the bottom of this email. We are now extending the project to staircases. Basically, we are implementing the method of limit together with the subsequent modifications (i.e., simple and transformed up-down) in a MLP like fashion. If there is any one out there that is familiar with these procedures, we would be happy to ask suggestions for setting the parameters of the experiments we will implement (e.g., number of reversal and step size). For sure we will implement all experiments that are now implemented in MLP. And the future extension of the project (if we make it) will be PEST... All the best, m & Alessandro Soranzo %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% (Please note that in the following text the word "level" refers to a any physical parameter you are manipulating with the maximum likelihood procedure) The two main modifications are: [1] it is possible now to repeat the first trial until the subject does not return a correct response. This is a pragmatic addition. The maximum likelihood procedure mistrack (largely) the subject's threshold if the subjects lapses an easy trial. Because the first trial is usually the easiest, a lapse on the first trial returns a very bad threshold estimation. If the "repeat 1th trial" option is enabled, the first trial is repeated until the subject does not return a positive/correct answer. [2] you can now edit the "level of the standard stimulus" within the graphical interface. Case 1. Let suppose the experiment is a frequency discrimination threshold for a 1-kHz pure tone. The frequency of the standard tone (i.e., 1-kHz) can be now set through the interface. In other words, if you want to run the same experiment with a 2-kHz standard tone you don't need any more open the experiment.m file but you can do it directly through the interface. Case 2. In case of detection experiments. If the experiment is a detection using a yes/no procedure the "standard level" becomes the "catch trial level". In other words, if you decide to use catch trials, when a catch trial is presented it will be presented the "level" of the stimulus written in the "standard (or catch) stimulus level" field. Case 3. There is a case 3, i.e., an alternative way to use the "standard level" field. However, this requires a bit of understanding of the program so please ask us directly. In any case, the meaning of the "standard level" is written in each experiment (type help experimentname) as well as it is visualised when the experiment is selected at the top of the graphical interface.

This message came from the mail archive
maintained by:
DAn Ellis <>
Electrical Engineering Dept., Columbia University