Can we learn without awareness ?

Listen to audio

Subliminal priming effects are generally brief, vanishing in a few hundred milliseconds, leading Lionel Naccache to propose a simple, if not caricatural formula: "The unconscious is structured like a decreasing exponential". Indeed, an exponential decay of priming effects is attested in various paradigms of subliminal or preconscious presentation. However, the rapid disappearance of non-conscious representations from working memory or iconic memory does not rule out the possibility that they may lead to longer-lasting modifications of the circuits concerned, for example in the form of changes in synaptic weights.

So can a non-conscious stimulus lead to lasting learning? We have deliberately left out of this discussion the vast field of implicit learning, in which participants are indeed aware of stimuli, even if they appear to be unaware of the temporal or spatial relationships associating them. In the experimental literature on subliminal perception, there are very few results in favor of learning without awareness. However, some recent experiments by Takeo Watanabe and colleagues suggest that massive repetition of stimuli presented below the threshold of consciousness can indeed lead to perceptual learning (Tsushima, Seitz & Watanabe, 2008; Watanabe, Nanez & Sasaki, 2001). In these elegant experiments, a stimulus consisting of multiple moving dots is biased so that a particular direction of motion is presented more frequently than the others. When only 5% of the dots are in coherent motion, this bias is undetectable, even after thousands of trials. However, repeated exposure of subjects to such a stimulus leads to a significant improvement in detection and categorization performance, and only for the direction that has been trained in this way.

Based on variants of this paradigm, Seitz and Watanabe (2005) put forward a theory according to which the non-conscious stimulus benefits from the concomitant presentation of a conscious reinforcement signal. Surprisingly, learning of a task-irrelevant stimulus is better when the stimulus is not consciously perceived than when it crosses the threshold of consciousness - presumably because, in the latter case, the subject triggers active strategic inhibition processes.

Further experiments by Matthias Pessiglione (2007, 2008) clarify the conditions for non-conscious learning. A masked visual stimulus can lead to operant conditioning, biasing motor responses, if paired with conscious reinforcement. Conversely, a masked motivational cue presented prior to trial, even if it remains non-conscious, modulates the conscious performance of a motor effort task. In all these experiments, however, the subject always makes a conscious effort, and either the stimulus or the reinforcement signal is clearly perceptible. As yet, there is no evidence of non-conscious learning taking place entirely without the subject's knowledge. In short, the forms of subliminal learning that have been identified so far bear no resemblance to the naïve vision of effortless learning through simple passive exposure to stimuli below the threshold of consciousness, or even during sleep - all these ideas are likely to be classified as "neuro-myths"!