Sign up Calendar Latest Topics
 
 
 


Reply
  Author   Comment  
aboharbf

Member
Registered:
Posts: 97
Reply with quote  #1 
Hey all,

Just switched over to the new runtime, and it has led to ISI of 500 ms being exceed much more often than before. attached is my code. Let me know if you see anything that would lead to problems. the system is only 2 years old. 

 
Attached Files
m fixate_reward_v2.m (4.20 KB, 2 views)

0
Jaewon

Administrator
Registered:
Posts: 954
Reply with quote  #2 
What do you mean by ISI? Are you talking about ITI? If so, please set the ITI to 0 and see how long your ITIs are usually.

If you are using movies, consider pre-creating them so that they don't need to be created every trial. It decreases ITI significantly. There is an example in the "task\userloop\2 movie preloading" directory.
0
aboharbf

Member
Registered:
Posts: 97
Reply with quote  #3 
I did mean ITI, sorry about that. Does the persistent variable require the task be wrapped into a function, as the movie_userloop trial is, or is it adequate to place code like this at the beginning of your timing file with the appropriate check to see if it has been created before?
0
Jaewon

Administrator
Registered:
Posts: 954
Reply with quote  #4 
It has to be a function separate from the timing script. The userloop function is something that replaces the conditions file. There is a detailed description about the userloop, if you read the "task\userloop\1 dms with userloop\dms_userloop.m" file
0
aboharbf

Member
Registered:
Posts: 97
Reply with quote  #5 
So, now that I've adapted this structure, things have worked out well. I'm now attempting to introduce a different paradigm - Rapid Serial Presentation. Essentially 200 On 200 Off for small images, continuously, with a reward every 3-4 images. I have the reward frequency worked out, but maintaining the fixation dot for as high a fraction of the presentation as possible seems like it will require me putting ITI to 0. Oddly, despite using images and preloading them, I get higher ITI limits than with movies. What can I do to bring ITI as low as possible?
0
Jaewon

Administrator
Registered:
Posts: 954
Reply with quote  #6 
In programming, even where you put a period matters. Unfortunately some of your sentences do not make sense to me. Please explain what you did again. And send me all your code and stimuli.
0
aboharbf

Member
Registered:
Posts: 97
Reply with quote  #7 
Apologies for the confusion. PM sent. basically I am looking to do a task where a trial is 200 ms of fixation followed by 200 ms of stimuli. In the current structure, the reward scene is making things a bit longer but I would like to find a way to drive ITI as low as possible.
0
Jaewon

Administrator
Registered:
Posts: 954
Reply with quote  #8 
I don't see any practical difference between the timing files you sent today and the day before. Are you sure this is the right file?

Your new description doesn't make sense either. What does ITI have anything to do with presenting images? Why is it necessary to reduce ITI? I must say to you the same thing that I said to others. You had better not just imagine the cause of a problem and ask how to do something to fix the problem as you want. Most people do not see things as programmers do. The better way is simply to describe the goal that you want to achieve in detail.
0
aboharbf

Member
Registered:
Posts: 97
Reply with quote  #9 
The script is basically the same (plus the fix you suggested), with stimulus presentation set to 200, and the fixation period set to 200, and the reward delay set to 1.

My goal is as follows: Have a task where 200 ms of fixation is followed by 200 ms of stimulus presentation (in this case, an image) with the fixation dot on top of it. This will immediately be followed by a new trial. This is typically called 200 On 200 off. After 3 - 4 successful trials, a reward is delivered (hence the conditional statement around goodmonkey).

In terms of examples which might provide clues as to how to implement this, the LBC fMRI example is another potential route for implementing this, if that gives you any closer an idea.
0
Jaewon

Administrator
Registered:
Posts: 954
Reply with quote  #10 
If you present only one image, why is it "rapid serial presentation"?
0
aboharbf

Member
Registered:
Posts: 97
Reply with quote  #11 
Wouldn't 2 images at a time be parallel?

Rapid = 200 ms
Serial = 1 at a time
0
Jaewon

Administrator
Registered:
Posts: 954
Reply with quote  #12 
Yes, but, if you present only one image, that is neither parallel nor serial.

And what you described above is not your goal. Your goal is what you want to do by presenting images in that way, like "I want to do test receptive field mapping in V1. To do that, I need to present 1000 images rapidly at 20-ms intervals...."
0
aboharbf

Member
Registered:
Posts: 97
Reply with quote  #13 
It is serial because runs aren't 1 trial. What the subject sees can be appropriately called 'Rapid Serial Presentation'.

My goal programatically is the same regardless of whether I am testing receptive fields for individual neurons in V1 or analyzing LFPs in MT. This is a common paradigm for those studying Inferior temporal cortex. 
0
Jaewon

Administrator
Registered:
Posts: 954
Reply with quote  #14 
I am a neuroscience Ph.D. and my job is to study the functions of orbitofrontal cortex, not to answer programming questions. If you want to talk with a programmer, find someone else.

The LBC fMRI example is what I made so that you can study it and make something on your own. If you think it does what you want, study it hard first.

I am not a mind reader and you don't have a telepathic ability either. Making yourself understood is the first step to get what you want. LBC people didn't mind spending hours with me to explain what they needed to do. Why should I expect less, especially when I do this as a favor?
0
Jaewon

Administrator
Registered:
Posts: 954
Reply with quote  #15 
Now it makes a little more sense. You think you can run a series of 400-ms trials with NIMH ML. It won't work. You should complete presentation in one trial.

It is important for you to give me a big picture of your study. When I worked with LBC people, I had to redesign things from the beginning a couple of time because it turned out that the total size of the image files they needed to present was up to several gigabytes. If they had given me better information, their experiment would not be delayed.

You sent me only 4 image files. Are those all the images you will present?
0
aboharbf

Member
Registered:
Posts: 97
Reply with quote  #16 
I know you're a PhD, as we met at Neuroscience 2017. Apologies for any misunderstanding, I greatly appreciate all the help you've provided over the years. I don't know what else I can say about the paradigm which I haven't already. This paradigm is a staple in the lab I work in, and my hope is to develop it using MKL so others in the lab (who currently use a very inflexible proprietary code) might transition, and to have the option myself (though my core experiments are with movies). I'm happy to describe things in as much detail as necessary. If there are any gaps in what I've described so far, please point them out.

Edit: Ok, I was hoping 400 ms trials would be possible. I will see how to arrange things in 1 trial. can eventmarker(9) be used to make it appear to be a new trial to machines recording the data? Otherwise a rather extensive re-write of analysis could would be required. The big picture of the experiment is simply to have 200 ms of baseline followed by 200 ms of stimuli, all the while having the fixation dot present to aid in maintaining the subject's fixation. I'll let you know if I run into any other issues. Thank you again for the help.
0
Jaewon

Administrator
Registered:
Posts: 954
Reply with quote  #17 
The eventmaker 9 (and 18) is a tradition from the original ML, but you can turn it off or use it for other events.

There are many missing pieces of information here. For example, you said that the fixation dot is there to aid the fixation maintenance. Then, is the fixation required to continue the trial or is it okay for subjects not to look at the dot? If the fixation is a requirement, when did the subject start it? Does the image presentation begin after fixation is acquired? You mentioned how images should be presented but nothing about behavioral requirements.

It would be great if you have a figure or diagram that shows the trial structure of the task, like the one you would put it in the methods of your paper.

And how many images do you need to present?
0
aboharbf

Member
Registered:
Posts: 97
Reply with quote  #18 
I meant using it in the middle of the trial to mimic shorter trials, on the Blackrock/Analysis side of things. Hopefully that does the trick.

Sorry about that. the reward would be contingent on fixation. the typical paradigm in the lab involves this 200 on 200 off image series, with fixation only because used to decide reward. If the subject chooses to never fixate, the stimuli will continue. so from the task perspective, it is ok if the subject doesn't look at the dot, they simply won't be rewarded. I would likely impose that the task begin upon the initial fixation, but continue regardless. It may be wise to make it so the task ends after a certain number of stimuli presentations without fixation, but that is not the current paradigm. 

I will draw it up at some point, but I think I'll end up leaving things out much like I did with the writing. Image presentation will be in the ballpark of 20 - 60, with the aim of repeating each at least 10 times, ideally 15 - 20, so in total, 1200 presentations. 
0
Jaewon

Administrator
Registered:
Posts: 954
Reply with quote  #19 
It sounds very much like the LBC fMRI example. What is the randomization strategy of those images? What if the subject blinks during the image presentation rather than actively breaks eye fixation?
0
aboharbf

Member
Registered:
Posts: 97
Reply with quote  #20 
Randomization would like be either pure or 'pseudoblocks' with all the stimuli with a particular tag are shown first, then the remainder are shown. 

Blinking ideally wouldn't be punished. 
0
Jaewon

Administrator
Registered:
Posts: 954
Reply with quote  #21 
See the LBC fMRI example (also ImageChanger). It already does everything you said, except the image presentation is triggered by the TR signal, instead of eye fixation.
0
Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.