Anyway to compute subject performance over the last 25 trials?

Support
1

Hi Chris,

We’d like to compute subject performance over the last 25 trials. In a typical programming language I’d create an array of length 25 and just fill in each slot with a 1 or 0 for performance, then cycle through the locations. I tried searching through Mworks to see if there was a way to do this, but I’ve been unsuccessful. I’d really like to keep this in Mworks, and not have to depend on Matlab or Python for it, any suggestions?

2

Hi Travis,

In a typical programming language I’d create an array of length 25 and just fill in each slot with a 1 or 0 for performance

You can do that in MWorks with a list. For example:

var performance = []

protocol {
    performance = []
    trial (nsamples = 25) {
        performance[size(performance)] = disc_rand(0, 1)
    }
    report ('performance: $performance')
}

The final report action produces a message like

performance: [1, 1, 0, 1, 0, 1, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 1, 0, 0, 0, 1, 1, 0]

You could also make the initial value of variable performance a list of length 25, but that isn’t necessary in this case.

Cheers,
Chris

3

Hi Chris,

Thanks for the quick response! Ok I’ll look into lists, I saw that in 2014
asmoly asked if lists could be updated (e.g. list[index] = 5), and your
answer was no, I assume that this is now possible? I’ve only used the
MWeditor to create and alter experiment xml’s so I’ll try to associate your
code to the editor’s steps. Do you have a link to how to incorporate code
like yours into the xml? I ask because in a previous question (I think it
was about using firmata with the manual reward window) you also sent code
as the example. Is it as simple as using an editor (ie sublime) and copying
the code into the xml? If this is the case, will subsequent MWeditor loads
automatically bring up the code? Sorry for all the questions.

4
Hi Chris, just wanted to shoot you a quick update and correction. I was confused in the earlier response in that there are 2 'Lists', one list is part of the Paradigm Components, and that's what I thought you were referring to, but I found a 2017 answer you gave that you updated the Dev version so that variables can be lists, and bingo, that was it! I have created a list P, with 25 default values [0,0...], and I'll have an index_counter that will cycle between 1 and 25, putting a 1 for correct, 0 for wrong. Then I'll just need to sum them up and divide into 25 for performance! I'll let you know the result, if you have any tips, feel free to share.

Thx!

T

5

Hi Chris,

It works!!! Some things I learned:
list variables start at 0, so if I wanted to go from 1:25, I actually
needed a list size of 26 zeros for the default value. To sum I just did,
P[1]+P[2]… and put that into a different variable that I then compute
performance on. Thanks for the help, and being a sounding board :wink:

Here’s how I set it up in MWeditor terms (for anyone else that wants to do
something like this):
variable P: type list: default value: 26 zeros.
variable i_P: default 1, this is a cycling index counter that goes from 1
to 25 then back to 1 …
When a trial is correct I assign P[i_P] = 1, wrong: P[i_P] = 0.
At the end I then assign a different variable var = P[1]+P[2]…
To compute performance I assign variable Performance to (var / 25)*100.

T

6

Hi Travis,

Glad you were able to get this working. My only comment is to discourage you from trying to force 1-based indexing onto MWorks’ zero-based lists. It might make things seem simpler at first, but ultimately it will make your code harder to understand and more prone to errors. (If you still want i_P to run from 1 to 25, just use i_P-1 as the indexing expression.)

Cheers,
Chris