In response to our early query regarding techniques for subliminal presentations, we got many good suggestions which are summarized below.
Thanks,
Moshe Bar
******************************************************
Date: Tue, 31 Oct 1995 16:20:26 -0800 (PST)
From: Moshe Bar <bar@selforg.usc.edu>
To: cvnet@skivs.ski.org
Cc:
Subject: Subliminal exposures
We want to display line drawings for subthreshold durations (5-20 msec).
Using the hardware we have (Macintosh Quadra 950), we are limited by the screen refresh time (~14 msec).
Does anybody have any experience with such brief presentations and could recommend specific hardware/software ??
Thanks,
Moshe Bar
**************************
Dear Moshe:
I have had limited experience with this problem. I'm sure you will hear more
>from people like Eli Pelli (address shown below if you want to contact him).
However, I know it can be done with appropriate coding.
I'd appreciate a copy of any responses that look interesting to you.
Thanks,
Robert Savoy
savoy@rowland.org
The text below is from 1993, so may be outdated.
ideoToolbox: Read me
January 18, 1993
Denis Pelli
Institute for Sensory Research
Syracuse University
Merrill Lane
Syracuse, NY 13244-5290
denis_pelli@isr.syr.edu
(Should be "VideoToolbox", which is Denis's name for the software collection
he and others have put together for doing vision research with Macs.)
***********************************
Remember that the display duration is NOT the relevant parameter but the
product of duration and contrast (Block's Law). Thus 15 ms may be below
threshold or above threshold, depending on the contrast you have.
H. Irtel
Prof. Dr. Hans Irtel
Url http://www.uni-mannheim.de/unima/fakul/psycho/irtel
Current Address: Zentrum fuer interdisziplinaere Forschung
Universitaet Bielefeld
Wellenberg 1
D-33615 Bielefeld
******************************
One way to do this is to use an oscilloscope connected to a board
that generates x, y, and z (intensity) signals. The Picasso
(John Daugman), or Cambridge Instrument display hardware can
display up to 200 Hz.
Another way:
With most standard VGA cards (i.e. for IBM-PC type machines) it
is easy to set a higher refresh rate, e.g. 200 Hz can readily be
achieved. This is done by reducing the frame size so that fewer
pixels are displayed (e.g. 512x512 instead of 1024x768). By
setting a smaller horizontal line size a higher horizontal
frequency is obtained, and by setting a smaller vertical frame
size, a higher vertical (refresh) rate is obtained. The smaller
line and frame sizes are achieved by setting the 4 horizontal and
4 vertical clock registers which are part of any standard VGA
card. This is easy to do with I/O functions available with any
C compiler library.
It is possible to overrate most VGA monitors to display
higher vertical rates than they are specified for.
For example it is possible to go up to 110 Hz for a NEC 3fg.
Recently, monitors spec'ed up to 150 Hz are being made
(e.g. NEC), so I would imagine that these might be overrated
up to 200 Hz as well.
So by combining a faster frame rate on the VGA with a high-speed
monitor, it ought to be possible to achieve 5 msec display
duration for one frame. The trick is to find a monitor that will
display correctly for both horizontal and vertical frequencies
that are generated by this method. By varying the sync duration
and timing it is often possible to get a specific set of
horizontal and vertical frequencies to sync properly.
If you are limited to Mac hardware, I would imagine that a
similar method might apply or that some vendor might make a
display board similar to the ones available for the PC type
machines.
Hope this helps,
Rob Smith
***********************************************
Dear Moshe
Our experience with animal trials shows that an effective and
inexpensive way of ensuring brief presentation times for stimuli using
a screen with a slow refresh time is to place the subject on a
rotating chair or stereotactic assembly, and place eye blinkers like
those commonly used to avoid peripheral stimulation of racehorses
during horse races. The effective presentation time of the stimulus
can then easily be adjusted by varying the angular velocity of the
chair.
Keep in touch,
Daniel.
**************************************
Dear Moshe,
We have just been discussing bringing out a new product that would
probably meet your needs. It is an add-on package for a PC that provides
a 21" monitor with a 3 ms frame time and ultra-short persistence
phosphor, a PCI graphics controller, and T-Scope software. The software
loads up to 8 PCX image files (probably 640 x 480 x 8 bit resolution) into
the board's VRAM and allows them to be sequenced with frame-time
accuracy. We plan on initially producing a DOS/Windows version. We
would also like to do a Mac version for Macs that can accept PCI
video cards, but that would come later - we aren't Mac programmers,
and it would be nice to see if we have a market! The price is in
discussion but would
be in the $5-6.5K range, depending on whether we decide to sell direct or go
through distributors (who take at least 30%.)
Because of the high framerate requirements for TScopes, proprietary
hardware is required - standard monitors won't do the job. Even
our standard research system, VisionWorks, has a 7.5 ms frame time
which is not really in the TScope range.
regards,
Steve
======================================================================
Steve Panish Ph.D. Voice: (603)-868-2090
Vision Research Graphics Fax: (603)-868-1352
99A Madbury Road E-mail: scp@curtech.com
Durham, New Hampshire, USA 03824
======================================================================
**************************************
Moshe,
A trick is possible which depends on vertical refresh rate and
vertical sync doubling. Using a high bandwidth multi-sync monitor
such as the Sony 17se1, you can present a subfield image for
75Hz * 2 = 150Hz --> 6.7 ms
72Hz * 2 = 144Hz --> 6.9 ms
67Hz * 2 = 134Hz --> 7.5 ms
60Hz * 2 = 120Hz --> 8.3 ms
and a normal image for
75Hz --> 13.3 ms
72Hz --> 13.8 ms
67Hz --> 14.9 ms
60Hz --> 16.7 ms
Clearly there are some limitations here but it's a cheap trick.
(A few other frequencies are possible with other display boards.)
The vertical sync doubling circuit is doable (StereoGraphics
CrytalEyes-PC does it for instance).
Gregg
Gregg_Podnar@SENSOR.RI.CMU.EDU
************************
Moshe Bar,
There is a package, macglib, made by a guy at P Cavanagh's lab
at Harvard that should do the job and many others for you. You have to
buy it though.
Also, there is a MacPsych list that
helps you answer specific stuff- I've
found most of the conversation there
relevant and helpful...
I was trying to find you more specific info, but failed. You can
contact with Kavanagh directly.
Good luck.
-- Zili
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Zili Liu, Ph.D.
Physical & Computer Science Research fax: (609) 951-2482
NEC Research Institute, Inc. voice: (609) 951-2734
4 Independence Way email: zliu@research.nj.nec.com
Princeton, NJ 08540 http://www.neci.nj.nec.com/homepages/zliu.html
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
*********************************
I suggest you follow the stimulus with a patterned mask.
This should be high contrast, long duration (200 + msec), and
patterned in such a manner as to incorportate many of the visual features
of the possible target stimuli.
Marcel (at MRC, Cambridge UK) has shown subliminal perception
using masking. Eg an 'inviisble' word such as Doctor will prime Nurse,
almost as well as if the 'Doctor' had not been masked at all !
Masking has some great advantages. The target, though brief, can be
shown at full contrast and therefore gets into the cortex. In the
usual procedure, stimuli are often invisible because they don't make
it past the retina (the contrast is so low, and on particular trials
so much of a target may not register at the retinal level that the
scrpas of info. theat get to the cortex are useless). So 'subliminal'
perception means 'no' perception on many trials - and these trials cannot
be distinguished from the useful ones. So many trials are wasted and
the effects are weak. Masking is a much better strategy.
..
Adam Reeves
***************************
Hi - in my work on 2-flash apparent motion, I have had some experience with
this. Before working on screen refresh rate, the most important trick is to
be sure and use light or dark stimuli superimposed on a uniform grey background
(rather than light stimuli on a black background). This will help minimize
if not eliminate effects of phosphor persistence.
There are various monitors which can run at higher than ca 60 Hz.
I use a NEC XP-17 at 160 Hz, for example; I think several current multi-sync
monitors can go up to 160. The Joyce display screen can to to 240, and has
an exceptionally short phosphor presistence, though I it suffers from more
frequent break-down than most others monitors, and is very expensive.
If you are using a Mac, your biggest problem is to get your graphics
card to put out a signal at a higher frame rate; normally, you are a
prisoner of the card maker's driver. You could write your own driver,in
principle, but you would need detailed technical info on your particular
graphics card, which is often hard to come by.
Denis Pelli says that Apple has agreed to release a graphics
driver control panel which lets you set any frame rate you want; but it
will work only for the on-board graphics controllers on only some of the
current Mac's (7200, 7500, and 8500, I believe). I have not heard if it
has yet been released.
Best wishes, hope this helps.
Curtis Baker
*******************************
hello there! I saw your query on CVNet for help on subliminal
exposures. You have probably already received a bunch of advice, so I
will only offer two words in outline form (one is even just a p.s.
technically.!.):
It turns out that, within certain limits, for exposure durations under
about 80 msec, luminance and presentation time can trade off in something
like a linear fashion. -- SOOoo, you can simply use "dimmer" drawings
presented at 15 msec (most monitors use 15 msec refresh). There is
research on this stuff so if you want more info/a reference, then feel
free to write me.
-james
p.s. I am not sure how you are planning on presenting the line drawings
(are you using some commercial software or programming in C..etc?). BUT,
as you may know, timing for full screen images is hard to keep precise.
So, one technique that is often used is a variant of lookup-table
animation. Roughly: 1) set the color lookup-table to all black. 2) load
the image to the screen (it will not appear since the lookup-table is
black) 3) change the lookup-table to contain the "right" luminances. 4)
change it back to black when presentation time is up. This method
involves changing something like 255 values (one for each lookup color)
instead of changing something like 300,000 values (all pixels on the
screen); it is therefore more likely that you can keep timings reigned in.
hope this was helpful -> and sorry if I presented stuff that was
totally obvious to you: it is hard to judge what level of info may
be useful.
======================================================================
James Intriligator
ji@wjh.harvard.edu
Vision Science Lab
Harvard University 617-495-3884, 617-643-0994
======================================================================
****************************************************
Not to worry -- Though your raster repeat time is 14msec, any small region
of your screen is scanned in a few 10s of microsec, and the phosphor
persistence on most screens is about 2-5 msec if you run not too far above
threshold.
Bruce Bridgeman
***************************************8
The "point plotter" from Edmonton/Canada is the only hardware I can
think of that would allow presentation times as low as you want.
Bernhard Treutwein here in our lab has (a lot) of experience with
it: bernhard@groucho.imp.med.uni-muenchen.de
Hans Strasburger
~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~-~
Hans Strasburger | hans@groucho.imp.med.uni-muenchen.de
Inst. f. Medizinische Psychologie | Tel: (+49)(89) 5996-217
Goethestr. 31 | Fax: -615
D-80336 Muenchen, Germany | priv: (+49)(89) 48 59 19
~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~--~~-~
***************************************
I have used a Uniblitz shutter for 10-60 ms presentations. LED's are the
best, perhaps in a "Maxwellian view" configuration, illuminating a
transparency of your test figures. The fact that each frame is scanned from
top to bottom in a computer monitor setup means you will have great
difficulty (impossible I beleive) getting even s8imultaneous presentation of
an image for 5 ms. The best I gave is a Cambridge Research systems card
which has a frame rate of 100 Hz, but the resolution is poor then. Try an do
it with LED's and transparencies if you can.
Phil Anderton
Optometry, Uni of NSW
********************************
I'm currently using a system (HW and SW) with 100 hz refresh rate ( 10 =
ms pr. screen) and 30 bit color depth. If this sounds interesting, =
notefy me and I will send you some more info.
Sincerely,
Lars Rune Bjoernevik
Huseby National Resource Center, Oslo, Norway
--------------------------------------------
Tel +47 22 02 95 00
Direct +47 22 02 96 09 Office
Direct +47 22 02 96 05 Lab
Fax +47 22 92 15 90
Internet: lrb@wuf.winhlp.no
msn : lrb@msn.com =09
CompuSrv: 100103,2144
Time Zone: GMT+1h
--------------------------------------------
*************************************************8
Moshe Bar
Psychology
University of Southern California
Dear Moshe Bar
I see from your recent posting on CV-net that you want to test subliminal
exposure by displaying line drawings on a CRT for short durations, so I'm
replying to introduce our range of tools for vison research.
Cambridge Research Systems equipment is designed specifically for vision
science applications. We apply the latest technology and tailor our systems,
making it easy to produce visual stimuli which would otherwise be difficult, or
sometimes impossible, to display.
Our main system is the VSG2/3 visual stimulus generator, a sophisticated
framestore card with many unique hardware features. The VSG2/3 can be configured
to drive virtually any monitor, and the system can be gamma corrected using the
OptiCAL photometer. Various options and peripheral equipment are available to
customise the VSG2/3 for specific applications; these include additional video
memory, an intelligent waveform generator, and shutter goggles. The VSG2/3 is
programmed using a library of high level functions which have also been
specially created and optimised. Standard programming languages are supported,
and many stimuli can be created in just a few lines.
I think that our system will be suitable for your particular application since
the VSG2/3 can be configured to produce the exact timings that are necessary to
display your short duration stimuli. I will be sending you our latest brochure
and set of notes for a variety of applications to give you a complete overview
of all our present systems.
We also have a web site that contains all the latest information about the
VSG2/3 and VSG Software so you can have a look and see if you think that the VSG
may be suitable for your particular application; you can connect to our home
page using the URL in the email signature.
I hope that you will find all the information contained in our brochure and
on the web site of interest and will contact me shortly with your questions,
problems and ideas.
Best regards,
Steve Elliott
-- Cambridge Research Systems Ltd. Tel +44 (0)1634 720 707 Fax +44 (0)1634 720 719 http://www.crsltd.com (last updated 18th October, 1995)