Effects of predictability and error on corticospinal excitability during an electronic prediction game

Background: Detecting environmental regularities is a sophisticated cognitive skill, but its neural basis remains unclear. We tested whether corticospinal excitability tracks stimulus predictability, prediction errors, and task progression while the participants played the Goalkeeper Game. Methods: Sixteen right-handed males (24.4 +/- 6.8 years) saved penalty kicks using index, middle, or ring fingers (left, center, right). Kick sequences were generated by a context tree model, in which contexts represent the minimum recent history required to predict the next event. Participants completed 1,200 trials (6 blocks x 200 trials each). Transcranial magnetic stimulation (TMS) delivered 400 ms before the go signal elicited motor evoked potentials (MEPs) in the first dorsal interosseous (FDI) and flexor digitorum superficialis (FDS) muscles during blocks 2, 4, and 6. Contexts were classified as unpredictable (1, 10) or predictable (2, 20, 00); prediction errors were defined for non-deterministic transitions after context 1. Log-MEPs were analyzed using mixed-effects models. Results: In FDI, a Predictability x Previous-Error x Block interaction emerged (F(2,8997)=6.89, p=.001), confined to the final block. Predictability increased MEP amplitudes after successful transitions (+6.3%; p=0.006) but decreased MEP amplitudes after failed transitions (-7.0%; p=0.007). For FDS, unpredictable trials elicited larger MEPs (+7.2%; p=0.002), and MEPs increased from block 2 to 6 (+16.8%; p=0.003). Conclusions: FDS MEP responses were broadly associated with uncertainty and task progression, whereas FDI MEP responses showed context-specific, error-dependent modulation after extended exposure, suggesting fine-grained predictive coding in goal-executing effectors.