Subliminal learning achieved notoriety back in 1957, when James Vicary claimed moviegoers could be induced to buy popcorn and Coca-Cola through the use of messages that flashed on the screen too quickly to be seen. The claim was later shown to be false, but though the idea that people can be brainwashed by the use of such techniques has been disproven (there was quite a bit of hysteria about the notion at the time), that doesn’t mean the idea of subliminal learning is crazy.
Ten years ago, researchers demonstrated that subliminal messages do indeed affect human cognition — and showed the limits of that influence . The study demonstrated that, to have an effect on a person’s decision, the subliminal message had to be received very very soon before that decision (a tenth of a second or less), and the person had to be forced to make the decision very quickly. Moreover, there was no memory trace detectable, indicating no permanent record was stored in memory.
But even such brief, low-level learning seems to require some level of attention. A study  found that subliminal learning doesn’t occur if the subliminal stimuli are presented during what has been termed an "attentional blink" You may recall when I’ve discussed multi-tasking, I’ve said that we can’t do two things at the same time — that tasks have to "queue" for attention. When a bottleneck occurs in the system, this attentional "blink" occurs.
But low-level sensory processing, which is an automatic process, isn’t affected by the attentional blink, so the finding that subliminal learning is affected by the blink indicates that subliminal stimuli require some high-level cognitive processing.
This finding has been confirmed by other studies. One such study  also has implications for reading. Participants in the study were shown either words or pronounceable nonwords and asked to perform either a lexical task (to identify whether the word they saw was a real word or a nonsense word) or a pronunciation task on the words. Unbeknownst to the participants however, they had been first presented with a subliminal word that either matched or didn't match the target word. People performed the tasks faster when the subliminal word was identical to the target word. However (and this is the interesting bit), the researchers then applied a magnetic pulse (transcranial magnetic stimulation) to the key brain regions of the brain before presenting the subliminal information. By applying TMS to one brain area or the other, they found they could selectively disrupt the subliminal effect for either the lexical or pronunciation task. In other words, it seems that, even when the stimulus is subliminal, the brain takes into account the conscious task instructions. Our expectations shape our processing of subliminal stimuli.
Another study  suggests that motivation is important, and also, perhaps, that some stimuli are more suitable than others. The study found that thirsty people could be encouraged to drink more, and also pay more for their drink, after being exposed to subliminal smiling faces. Subliminal frowning faces had the opposite effect. However, how much, and whether, the faces had an effect on drinking, depended on the person’s thirst. Those who weren’t thirsty weren’t affected at all. Smiles and frowns are of course stimuli to which we are very responsive.
So clearly, although it is possible to be unconsciously affected by stimuli that can’t be consciously detected, the effect is both small and fleeting. However, that doesn’t mean more long-term effects can’t be experienced as a result of information we’re not conscious of.
Psychologists make a distinction between explicit memory and implicit memory. Explicit memory is what you’re using when you remember or recognize something — it’s what we tend to think of as "memory". Implicit memory, on the other hand, is a concept that reflects the fact that sometimes people act in ways that are clearly affected by earlier experiences they have had, even though they are not consciously recalling such experiences.
Another study  that used erotic images (because, like smiling and frowning faces, these are particularly potent stimuli, making it easy to see a response) found that when your eyes are presented with erotic images in a way that keeps you from becoming aware of them, your brain can still detect them — evidenced by the way people respond to the images according to their gender and sexual orientation.
The study is more evidence that the brain processes more visual information than we are conscious of — which is an important part in the process of determining what we’ll pay attention to. But the researchers believe that the information is probably destroyed at an early stage of processing — in other words, as with subliminal stimuli, there is probably no permanent record of the experience.
Which leads me to sleep learning. This was a big idea when I was young, in the science fiction I read — the idea that you could easily master new languages by being instructed while you were asleep.
Well, the question of whether learning can take place during sleep (and I’m not talking about the consolidation of learning that’s occurred earlier) is one that has been looked at in animal studies. It has been shown that simple forms of learning are indeed possible during sleep. However, the way in which associations are formed is clearly altered even for simple learning, and complex forms of learning do not appear to be possible.
As far as humans are concerned, the evidence is that any learning during sleep must occur during the lightest stage of sleep, when you still have some awareness of the world around you, and that what you are learning must be already familiar (presented previously while you were awake and paying attention) and not requiring any understanding.
All the evidence suggests that, although information can be processed without conscious awareness, there are severe limitations on that information. If you want to "know" something in the proper meaning of the word — be able to recall it, think about it — you need to actively engage with the information. No free lunches, I’m afraid!
But that doesn’t mean unconscious influences don’t have important implications for learning and memory. A paper provided online in the Scientific American Mind Matters blog describes how a single, 15-minute intervention erased almost half the racial achievement gap between African American and white students. And this is entirely consistent with a number of studies showing how our cognitive performance is affected by what we think of ourselves (which is affected by what others think of us).