Subliminal Perception---Old Hat or is There Something New?

There it was again and even more starkly put! "Researchers have found the first physiological evidence that subliminal images do registers subconsciously. Brain scans showed that even though volunteers were not aware of the series of quickly flashed images their visual-processing centers activated in response to the picture. The study from University College London did not address whether subliminal images could influence thought or behavior."

This time it appeared in Scientific American Mind, a relatively new spin off from Scientific American. A more complete PR blurb titled "Subliminal Advertising Leaves its Mark on the Brain" had come across my computer before. I read it quickly and considered putting it on the blog, since intuition may sometimes involve subliminal perception. It wasn't because I thought physiological evidence for subliminal perception was news. Rather it was for what appeared to be two almost contradictory reasons. It seemed an example of PR on overdrive and it provided an interesting summary of state of the art knowledge of subliminal perception.

I was under the impression that there was scads of evidence that subliminally presented images register on the brain. It seems to me that a full bibliography would take many pages, and that is probably an understatement. I’m not talking about subliminally presented pictures of popcorn at a movie theater increasing popcorn sales, nor flashing the RATS part of “Bureaucrats” in a separate frame in a Republican attack ad affecting votes. To the best of my knowledge such things have not been shown. This does not necessarily mean they cannot occur. (Related effects have been shown. For example, well-established work by Zajonc shows that picture flashed subliminally can influence subsequent preferences. The devil, as usual, must be in the details.)

Support for subliminal perception comes from brain imaging studies as well as behavioral studies. For example, a landmark paper published in 1998 demonstrated that presenting subjects with subliminal pictures of faces with fearful expressions activates their amygdala, the part of the brain that registers fear (Whalen et. al.) . A 2005 study went much further. It showed that threatening words that do not come to awareness nonetheless generate relatively long-lasting activity in the amygdala. Interestingly the amygdala did not begin to respond until about 800 milliseconds later (Naccache at al.). This delay makes sense when you think about it because words have to go through several levels of processing before their meaning is extracted.

Unconscious cognition has become a hot area of research. Researchers are interested in the topic for itself, and also because it is an approach to trying to understanding consciousness. To understand consciousness it helps to compare it to something. Often some kind of stimulus is presented that does not come to awareness ---for example because is flashed for too short a time. The brain activity is then compared to the brain activity when the stimulus is allowed to come to awareness. The difference is an indication of the neural activity associated with consciousness--- or the neural correlates of consciousness, as they are called.

It turns out that there are a number of ways to be sure visual images do not come to awareness---besides flashing them for too short a time. These techniques are called “masking.” In one technique a brief stimuli is presented and then an additional one is presented for roughly twice as long right afterwards. The second stimulus assures that the initial one does not come to consciousness. This is called backward masking. In another technique called the attention blink, attending to an initial conscious stimulus prevents another one presented from 200 to 800 milliseconds later from coming to consciousness. The University College of London experiment, by Bahrami et. al., used yet another technique. They presented faint pictures of tools to one eye and a continuous stream of high-contrast masks to the other eye, which suppresses conscious awareness of the tools.

This lab is very reputable, so I decided to read the paper. I am delighted I did. It confirmed something very subtle about the role of attention in subliminal perception---something well worth contemplating.

The researchers presented the faint images of tools under two different conditions. In the first, subjects were engaged in a relatively easy task---picking out a specific letter, such as a “T” from a continuous visual display of letter. In the second they were engaged in a more difficult task requiring more focused concentration---picking out a letter of a specific color, say a blue T. With the easy but not the hard task, activity associated with the tools showed up on the primary visual cortex. This suggests that information impinging on the retina at least under these conditions is only forwarded to the part of the brain that registers visual experience when the demands on attention are relatively light.

Researchers used to think that subliminal perception was independent of attention. Not only can attentional load interfere with sensory information coming to consciousness (check this out), it can also interfere with it even being recorded by the brain. Indeed the title of the paper---which unfortunately was not mentioned in either PR blurb---is "Attentional Load Modulates Response of Human Visual Cortex to Invisible Stimuli." (Other experiments had suggested this but had not proved it conclusively.)

How might the effect work? We have at least a partial answer. The brain is densely interconnected both within its different regions and between them. This allows it to integrate incoming information from the environment with other information. Sensory signals from the retina go through a relay station, in the thalamus---called the LGN, or lateral geniculate nucleus---on their way to the primary visual cortex in the back of the brain. It turns out that roughly 80% of the LGN's input is from higher brain centers. So, when demands on attention are high, subliminal sensory input somehow can get stopped at the LGN and will not advance to the visual cortex.

With respect to intuition one might be tempted to conclude that subliminal information only registers in the brain proper when it is not otherwise engaged in a demanding task. But is this really a fair conclusion? Let’s change the situation a little bit. For example, say faint pictures of enchiladas are flashed to hungry subjects who love Mexican food or frothing glasses of beer to thirsty beer-loving subjects ---under the same two conditions. Or how about instead flashing a definitive pictorial clue to an engaging and difficult insight problem the subject was given the night before. I have a strong hunch that at least the first two of these high valance, or emotionally charged, stimuli would compete more effectively with the ongoing attentional load and at least register in the visual center in brain. The third is a bit more complex. Especially here, I can hardly wait to see what happens.