Question about MultiTarget AcquiredTime

Hi all,

I am using 'MultiTarget' function to acquire choice reaction time.

However, I found the AcquiredTime in ChoiceHistory sometimes was even early than the event marker I set for run_scene.

For example, when I run the following code:

run_scene(scene8,Response_Start);

The event code time of 'Response_Start' is 4115.53, and the ChoiceHistory AcquiredTime is 4100. The RT output is 0.

Any idea why is this happening? Thanks!

It is because the behavior tracker (eye, joystick, etc.) was already at the target location before the scene started. In your case, the scene started at 4100 but the eye was already there, so the acquired time was 4100. The target graphics were turned on at 4100 as well, but the time when they actually showed up on the screen was one frame later, which was 4115.53. So the fixation (i.e., the choice) was not a response to the visual stimuli.

Hi Jaewon,

Thanks for your kindly reply. I'm still a little confused about the event time.

Since my task needs to make two sequential choice by eye, HoldTime for each choice is 400ms. I run two MultiTarget scene in a row and set a eventmarker between these two scenes.

run_scene(scene8,Response1_Start); %choice 1
eventmarker(Response1_End);
run_scene(scene9,Response2_Start); %choice 2

1. The time of 'Response1_End' is 50ms more than the Acquired time of first choice plus 400ms (HoldTime) on average. Which is the real time when the subject finish the first choice?

2. Is the second choice reaction time actually (choice 2 acquired time - choice 1 acquired time - HoldTime) in my case? I checked the choice 2 acquired time, and some of them were earlier than 'Response1_End' time. It's weird that the time difference between 'Response1_End' and ‘choice 2 acquired' in nearly 10% trials were less than 1 frame time, both positive and negative.

Thanks again!

Both cases look perfectly fine to me, but, to be sure, I need you to show me more than just the code. I need to know things like where the targets were, how the eye moved, etc. The best thing is to send me the data file and the task code.

1. What do you mean by "the time when the subject finished the first choice"? Are you asking when the HoldTime was over?

2. I do not understand how you come up with the reaction time expression. AcquiredTime indicates the time when the eye crossed the fixation window. It is calculated back in time once the fixation is confirmed. So it is normal that sometimes AcquiredTime of Choice 2 is earlier than Response1_End.

I got your message. I like your timeline figure, but just redrew it to show what might actually happen. There are a few things for you to know, to understand how MultiTarget (or SingleTarget) works.


MultiTarget (and SingleTarget) does not change its fixation status until the tracker remains in the target window for one entire frame. It means that, even though the eye crosses the threshold window sometime during Frame 0, the adapter does not think that the fixation is made until the end of Frame 1. So counting HoldTime starts from the beginning of Frame 2. The reason for doing this is to prevent a jitter from being detected as a fixation. It is also because the eye is still moving at the time of window crossing and the actual fixation does not start from that moment.

Choice2 AcquiredTime can be earlier than Response1_End, if the eye is already in the target window before the scene starts, as drawn in the above figure. In that case, the RT property of MultiTarget becomes 0, since the fixation is not a reaction to the targets. Note that the data collected during the last frame of Choice1 is analyzed at the beginning of Choice2.
https://monkeylogic.nimh.nih.gov/docs_CreatingTask.html#RuntimeVersion2_Background

The targets turned on during Choice1 are turned off at the end of Choice1 but the screen is not updated until the first frame of Choice2 (Response2_Start). So Response2_Start is the time when the subject sees the removal of the chosen target. The unchosen targets in Choice1 are turned on again before Response2_Start so that they appear to exist continuously.

The actual hold duration may not be exactly 400 ms. Please see this post for the details.

To answer your question, the reaction time for the second choice should be (Choice2 AcquiredTime - Response2_Start). The cue for the second choice in your task is that the target chosen from Choice1 disappears, which is timestamped by Response2_Start. The AcquiredTime sometimes being earlier is just because your subject makes estimated saccades before the chosen target disappears. You need to vary Choice1 HoldTime so that it is hard for the subject to guess. Or skip the reward if the AcquiredTime is earlier than Response2_Start.

Thanks so much for your detailed explanation! Really helped me a lot.