Attending is not enough: Responding to targets is needed for across-trial statistical learning.

It should be noted that while numerically the same effect was also present when contrasting predicted and neutral trials following Go trials [ β = −12.90, SE = 8.55, df = 59.90, t = −1.51, p = .14], it failed to reach significance, arguably to the low trial count in the neutral condition (~six trials).

also in 4,113 other papers