Awareness is necessary for predictive learning and prediction-based motor attenuation

Predictive coding theories posit that the brain continuously updates internal models of the world to anticipate sensory input. Attenuation of neural responses to predictable sensory stimuli is thought to aid in this process of prediction error detection and model updating. The topic of whether predictive processes such as sensory attenuation can occur in minimally conscious states has been hotly debated. The current study provides a novel test for the role of awareness in prediction-based attenuation using a transcranial magnetic stimulation (TMS) prediction task, whereby predictable stimulation is known to lead to attenuated motor system excitability. The experiment tested whether motor attenuation can occur without awareness using a dual task paradigm to mask a predictive cueing relationship. Participants simultaneously completed an n-back task and a TMS-counting task; critically, a visual cue (in the n-back task) consistently predicted the occurrence of a TMS pulse (in the counting task). Instruction conditions manipulated how much participants were informed of this relationship, and awareness was assessed in a post-experimental questionnaire. The results reveal that motor attenuation was present only in participants instructed about the relationship, or in those who became aware of the cue-TMS association. Implicit, predictable cue-TMS exposure was not sufficient to elicit motor attenuation in unaware participants. These findings show that awareness of the relationship is necessary for predictive learning and prediction-based attenuation in the motor system, challenging the assumption that predictive coding operates automatically.