highly significantp = 0.0004
We found that our decoder predicted the correct CA3 input pattern 71.4% of the time, which was highly significant compared to a shuffle distribution where CA3 input labels were randomly assigned (Figure 6e ; p = 0.0004; bootstrap test).