Why can I see my subliminal stimuli?

First thing to do is check out the section in the manual regarding data and the log file--and also the stuff on screen refresh rate in the introduction.

Also, take a look at the "time n" data columns produced in the log subfolder of the data folder. The log files will tell you after a session what [I]actually happened[/I]--as opposed to what you [I]asked for[/I] (e.g., if you requested that a picture be on screen for 5ms but it was actually shown for 16ms). This is a nice way of determining if your system can handle what you’re asking for. Note that if you are only displaying stimuli and not taking any RTs you may use negative block IDs (such as -1, -2, etc.) to force DirectRT to write the trial data to the log file.

With regard to refresh rate, remember that many systems will use a 60hz (16.7ms) refresh rate for many screen resolutions unless you manually set it while in that resolution--and after that Windows will remember it.

e.g., Let's say you are going to run a session in 800x600 mode (from "Options" in the Style file). You'll need to tell Windows what you want the frequency of the monitor to be for that specific resolution. By default it may be 60hz. To set it to a higher value, go to "Display Properties" in the control panel and set the resolution of the monitor to 800x600. *Then* go to "advanced" and set the refresh rate as high as you can for that resolution. You need to repeat this procedure for any resolutions you plan to run in because most PCs remember display settings separately for each resolution. e.g., if you set the frequency to 100hz for 1024x768, it may still run 800x600 at 60hz, etc.

Note also, that it is often easier to obtain faster and more reliable display speeds when you use lower resolutions such as 800x600, and when you are using a high quality graphics card on a system with lots of memory (e.g., 512mb+ is usually ok).

You should also take a look at the card driver. Sometimes an old or invalid driver is installed. This can also be responsible for seeing subliminal stimuli. If you have a card with on-board video-RAM (e.g. ATI Radeon or an older RageFury), normally you don´t have to upgrade your mainboard RAM (we use Elsa Erazors, 32MB onboard video RAM, 128MB mainboard-RAM). I´d say you should only take this into account if your videocard shares memory with your mainboard RAM.

See also a thread on CRT vs LCD monitors:
[URL="http://www.empirisoft.com/support/showthread.php?t=71"]www.empirisoft.com/support/showthread.php?t=71[/URL]

And also this post on timing discrepancies:
[URL="http://www.empirisoft.com/support/showthread.php?t=599"]www.empirisoft.com/support/showthread.php?t=599[/URL]

[I][edited from support email][/I]

I'm currently designing a subliminal priming study and am working on reducing the display time of the subliminal prime (as it can be seen as a flash on the screen). In Windows I have reduced the screen resolution to 800x600 from 1024x768 and raised the refresh rate to 85 hertz from 75 hertz. I then went into DirectRT and changed the resolution to 800x600, after which I ran the refresh rate test in DirectRT, and it indicated a refresh time of 11.77ms. I then lowered the display time for the subliminal prime to 12ms (from 17ms) in my input file. This is where my problem comes in. When I run the experiment in DirectRT, the screen resolution seems to revert back to 1024x768 at 75 hertz (the font of my text looks the same size on the screen in both resolutions, which it shouldn't). When I check the log file, I'm getting the exact same display times for the subliminal prime as I was before I changed all of the resolutions (around 17ms). I tried a few other resolutions and the same thing keeps happening. I have checked the help file and the forums, and did not seem to find anything that would help. Is there anything that can be done, or am I stuck with a display time of around 17ms? Any help would be greatly appreciated.

Hmm. Is your style file located in the same folder as your input file? If so, can you confirm that the display setting is set to [I]800x600[/I] when you click on [I]Options[/I]? If so, try creating two new images in your [I]stim[/I] folder--one 800x600 and the other 1024x768. Then try to display each using your current input file. Does the 1024x768 image fill the screen? Is there blank space surrounding the 800x600 image?

The style file was located in the same folder as my input file, and, using the method you proposed, I discovered the file was not being changed when I edited it through Edit > Edit Default Styles > Options in the DirectRT window. What I had to do was open up the style file directly from a Windows explorer window, and from there I was able to change the resoultion for the experiment, so now my display times for the experiment are lower. I am still seeing a flash on the screen, however, and have tried the fixes proposed here on the forum and in the help files, and I'm starting to think that I have reached the hardware limit of the system DirectRT is installed on, though I'm open to any other ideas.

The "default" styles file is the styles.drt file located in the main DirectRT folder. It is used when no style file is found in the folder that contains your input file. With regard to the flashing, it's entirely possible to "see" the flash of a 10ms display. What kind of display times are now being recorded in your log files?

When I was having the original problem, I did not realize that the folder which conained my input file also contained a style file. Modifying that file did fix my problem.
The log file is currently showing display times of between 12 and 14ms. The refresh rate test function in DirectRT gives me a refresh time of 11.77ms at a resolution 800X600 and a refresh rate of 85 hertz (these are my display settings in windows as well).

We're having difficulties with display times of subliminal primes in the beta version of DirectRT 2010. The words are set to display for various times, but the words are very visible and the log files show that they are displaying for far longer than that. Updating the video drivers did not help, nor did the advice that I've found here. I had no problems with older versions of DirectRT beta...Any ideas?

Jiff128, Ssmith,

What versions of DirectRT are you using, i.e., 20xx.x.x? In cases where it worked well using a prior version, was it the same same machine? If so, what version of DirectRT was that (if you can recall)? Also, if you run some of the various samples that come with DirectRT (especially ones from the first few folders), do you get similar variance in the display times?

We are currently using DirectRT 2010.122.804.10. We have run several priming studies on these same computers in the past using DirectRT 2006 (and an earlier beta version of DirectRT 2010 - last December), and the presentation times were much more precise (within 10-15 ms). The computers have 1.83GHz processors, 1GB RAM, DirectX 9, and the monitors are set to 1024x768 pixels at 85Hz.

We do get similar variance using the samples that come with DirectRT.

For instance, the sample in "4 - Applications\image priming" presents an image prime for 100ms. During a practice run the log file shows that these primes were presented for an average of 190ms instead of 100ms (SD = 44ms, ranging from 110 to 281ms).

As another example, the sample in "4 - Applications\standard priming" presents a prime word for 10ms. During a practice run, this prime was presented for an average of 114ms instead of 10ms (SD = 27ms, ranging from 49 to 159ms).

That is very unusual, I haven't heard of variance of that degree before. In your main DirectRT window, do you have an UPDATE option? If so, can you click to update your beta to the most recent version and see if the image priming sample gives you a similar pattern of data? I'll pass this on to Trevor our lead programmer who might have additional suggestions. Attaching a dxdiag.txt file could also help so we can see the details of your hardware (e.g., to figure out why v2006 could perform better on this machine).

I updated the beta version so it's now 2010.124.820.10, but the timings are still too long. With the images sample, the timings run from 574-596 (instead of 500) and 125-197 (instead of 100). With my own .csv file, I'm also seeing extremely long timings (63-172, though it's set for 35). I'm not sure what a dxdiag.txt file is...

I have my program running in windowed mode, though I have some colleagues who are using the full screen and have the same problems.

Is there any way we could talk directly tomorrow afternoon to try to get this sorted out? Trevor once helped me by logging into my computer and looking at the files I was using. Is that a possibility?

Here's the dxdiag.txt (I figured it out...)

SSmith,

I took a look at your dxdiag.txt file, thank you, and all seems normal. I am going to run through the same input files using the same version of the beta you are using. There is a more recent version of the package available and that may help but I will test both versions on one of our XP machines here and see if I can replicate the RT variances you are seeing. Fair to assume you did not change the resolution/style files in any way on the samples?

Blair

Thank you for your response. I will try updating to the newest version and get back to you. Please let me know what the outcomes are of your tests. And no, we did not change anything with the resolution or style files with the samples.

I tried to update DirectRT to a newer version of the beta, but it seems that the most recent update is the version we're already running (2010.124.820.10). Is there another way to get the update?

I know that the version of the beta that was available Nov/Dec 2009 did not have this problem on these computers b/c I was pilot testing a study with it and the primes weren't visible.

I'll await the results of your testing.

[I][edited by moderator] [/I]... my dept's computer person put the 2010 beta on a newer computer and the program is now working fine. I'm not sure what changed in the beta version from last Nov/Dec to now, but whatever it is does not allow it to work properly on older computers.

We have posted a new release that will require a full uninstall/reinstall. This is a pre-release of the final 2010 version--i.e., it's no longer a beta version. Here is a link to the November 11, 2010, pre-release:

[I][link[/I][I] removed][/I]

If you are reading this after November 15, 2010, you should check the [URL="http://www.empirisoft.com/download.aspx"]www.empirisoft.com/download.aspx[/URL] to see if a newer version of DirectRT 2010 has been posted. The post above refered to version v2010.2.102. It fixes some inconsistencies we were finding in the timing of screen displays of some system configurations. SSmith, it may resolve the issue you have on the older machines. Glad to hear the later beta works well on your newer machines. Feel free to check it out.