Background
Imagine the power of an advertiser who could put persuasive thoughts into your mind without your conscious awareness!  This is the idea behind so-called "subliminal" advertising.  According to some popular accounts, movie theaters can sneak frames into the move that flash, "eat popcorn," and the audience, supposedly unaware of having read this message, will comply and buy more popcorn.  The idea is that the message is displayed for such a short period of time that the viewer does not have time to consciously process it, and so is therefore not consciously aware of having ever seen the message.  According to proponents of this idea, the viewer will nonetheless process the message at some level below consciousness and thereby act upon it. 

At first blush, this notion sounds very possible.  After all, a great proportion of our mental processes and subsequent behaviors happen at a level that is below our level of awareness, even though these can be consciously controlled: breathing, eye blinking, scratching a nose.  Have you ever been in a crowded room at a noisy party?  Imagine that you are involved in a conversation with a few friends, totally unaware of what is being said in the conversations of dozens of groups around you.  Then you suddenly hear your name mentioned in a conversation three groups away, and your heart steps up its pace just a little as you wonder what they were saying about you!  Ahah - here is solid evidence that we do indeed process information at a level that is below our everyday levels of consciousness, and that this information is being monitored in a way that can immediately affect our conscious thoughts and behavior.  (This is called the coctail party phenomenon.)

However, if it really is possible to unconsciously influence people's thoughts and behaviors in adverising, the word would certainly be out and every advertiser would be doing it.  Consultants who know how to do this stuff would be making lots of money giving seminars on how to do it.   Indeed, why give seminars at all - why not just directly tap that unconscious part of our brain and directly feed the information into memory?  The reason is that the multiple ways that our brain processes information is not so simple, and we appear to have a lot of different kinds of processors and pre-processors.  For example, there is some pre-processing that filters out multiple conversations at a party as a way to keep our main processor (consciousness) from becoming overloaded with information.  The pre-processor is too small to hold very much, so it immediately flushes apparently useless information into a shredder.  Any useful piece of information, such as the mention of our name, is passed along, but we cannot recall any other contextual information, no matter how hard we try, because it has already been flushed and shredded.

We cannot find any practical evidence that "subliminal" advertising works, and yet we do know that pre-conscious processing systems exist.  The possibility that we could figure out how to tap these pre-conscious processes for financial gain or for more wholesome purposes, such as improved learning in school, makes this an intriguing area for study (albeit frustrating).  Below is one way that has been used to study this fascinating topic.

The Tachistoscope
A tachistoscope, or t-scope, is an istrument that can flash information before our eyes for very small and precise amounts of time.  There are two basic types: a flashtube tachistoscope and the shutter tachistoscope.  The flashtube t-scope uses a small lamp that is something of a cross between the fluorescent tube in ceiling lights and the flash tube used in a camera.  The flash tube in a t-scope has very fast "rise" and "fall" times in how fast it reaches maximum brightness and how fast it turns off, and it can be kept "on" for very precise amounts of time.  This makes it possible to light up an image for a few milliseconds at a time. 
 

FLASHTUBE TACHISTOSCOPE    The image cards (red, green, and blue) are individually lit by the flashtubes on the side.  The red image shines through a half-silvered mirror to reach the subject viewing through the left.  The blue image shines through a half-silvered mirror and reflects off of a second mirror to reach the subject.  The green image shines off of two mirrors.  Timing and sequencing is controlled with a computer.  (Click for interface details.)

A multi-channel flashtube t-scope uses mirrors in a large box so that when subject looks in the viewng hole, any of several images that might be lighted appear to be in the same place.  In a typical experiment, a "subliminal" image on a card is flashed for a few milliseconds, then a "mask" card is flashed for a few milliseconds longer (with the subject being unaware that the "subliminal" message ever appeared), and then some other final target image is flashed for a second or two.  The reason for using the "mask" image is to flush the "subliminal" image from the subject's retina.  Ever notice that you see lights long after a camera flash?  Since the retina in your eye can thus serve as a form of memory, and so we need to present another image to flush out the retina. (This is why it is likely that you would recall seeing an "eat popcorn" frame in a move if it had ever been done - as when you can clearly see the bubbles at the end of a film reel.)

In an experiment using a three-channel t-scope as described above, the subject would report seeing only the final target card, or perhaps noticing some sort of flash preceeding the target card.  In one experiment (Tony Greenwald), subjects were presented with affect laden words at the pre-conscious level of a few milliseconds.  These would be words that elicit some sort of emotional feelings, such as hate, love, fear, disgust, etc.  The final target word that was flashed at a conscious level was a meaningless nonsense word (e.g., "prind").  A computer was used to control the timing of presentations in the three-channel t-scope, and was also used to collect information regarding the subject's response to the final target word (and reaction time). 

Interestingly, the researchers found that the pre-conscious affect-laden word did appear to influence subject's feelings associated with the nonesense target word, even though no subjects believed that they had seen anything before the presentation of the target word.  That is, a person's feelings about a neutral object can indeed be influenced below the threshold of awareness.  However, the effects of such "subliminal" influence are very short lived.  Subjects who were asked to indicate the feelings associated with the nonesense words a few minutes later provided resultes that showed no correlation with the pairings that had been presented to them on the t-scope.

A problem with the flashtube t-scope is that it is very cumbersome to use.  With its mirrors and flashtubes, it is large, heavy, and fragile.  For the poor subject, it is somewhat uncomfortable to sit for some period of time with her/his head against the viewer of a large, immobile box.  Another instrument that can be used in a more natural way and with larger groups of subjects is the shutter t-scope.  Mechanical shutters, much like what is used in a camera, are placed in front of slide projectors.  A computer is used to cycle the projector slides and to open the shutters at the right times for the right amount of times.  Because the shutters are mechanical, they cannot be controlled with as fine a resolution and with the same precision as with a flashtube t-scope, but they are good enough for some applications.  Using the two-channel set-up shown below, it was possible to flash the iris-style shutters reliably down to around 10 milliseconds at the center of the iris. (Click for interface technical details. Click to zoom image of shutters.)
 

T-Scope Shutter Set-up Screen (TOP) T-Scope Shutters on Moving Cart (LEFT) (click on image to zoom for details)

The shutter t-scope above was used in one study to see if it was possible to influence a person's perceptions of a political candidate with pre-attentive information (Jon Krosnick).  An affect-laden word or object would be flashed briefly before showing the photograph of a political candidate.  In another experiment, the sequencing shown on the screen above was used to study the mere exposure effect (Tony Greenwald).  The idea of "mere exposure" is that people tend to form positive attitudes if they are merely exposed to something - e.g., if a person sees an advertisement several times that merely says, "vote for Smith," the person is likely to form a positive attitude about Smith where we would expect no rational reason to do so.  With the settings of the screen above, subjects were exposed briefly to such messages, followed by a mask (picture of a quilt) for two seconds, allowing the experimenter to study the influence of differences in exposure time on the mere exposure effect.