People tend to do what they are rewarded and avoid what they are punished. Given a rule for rewards (e.g. an extra bonus for hitting a red target rather than a green one in a game), over time people can learn to maximize the rewards by adjusting their behaviors and perceptions. This is the main idea of reward learning.

Since neuromodulatory signals for rewards and punishments are released diffusely throughout the entire brain, one hypothesis is that the effects of reward learning could be free from the constraints of consciousness. Unfortunately, this hypothesis has not been proved by the previous studies, in which perceptually distinguishable visual cues (e.g. red vs. green) are used to associate with different reward values.

A research team led by Dr. BAO Min from the Institute of Psychology of the Chinese Academy of Sciences (CAS) invented a novel paradigm to resolve this issue. The research team presented a target in one eye of the participant, and a dynamic sequence of geometric images in the other eye.

Usually, the target is rendered invisible due to the suppression from the dynamic images in the opposite eye. However, over time, the suppression will become ineffective, allowing the target to break into awareness. Participants were rewarded only when they reported seeing the target in one of the two eyes, which was called the rewarded eye.

Because the targets presented to both the rewarded eye and non-rewarded eye were of the same appearance except for their eye-of-origin information, participants could not consciously differentiate the monetary rewarding and non-rewarding targets.

Surprisingly, the targets in the rewarded eye still broke into awareness more than those in the non-rewarded eye during the training, demonstrating an eye-specific unconscious reward learning. The researchers further revealed that producing this effect required both top-down attention and inter-ocular suppression.

The study confirms that effects of reward learning can be free from the constraints of consciousness. Before being accepted for publishing, its preprint at bioRxiv has been cited by a recent review article published in Progress in Neurobiology which proposes a Predictive Global Neuronal Workspace (PGNW) model for visual consciousness. The article remarks that the above-mentioned finding of eye-specific unconscious reward learning supports the inclusion of a pragmatic component of their PGNW model.

The study entitled "Reward Produces Learning of a Consciously Inaccessible Feature" has been online published on June 22 in British Journal of Psychology.

This research was supported by the National Natural Science Foundation of China, the Strategic Priority Research Program and the Key Research Program of CAS, the Beijing Municipal Science & Technology Commission, and the Fundamental Research Funds for the Central Universities.