Fig. 1.

Experimental paradigm, intracranial EEG electrodes, and analysis approach. (A) A DMS task was used in the experiment. The same picture was paired with different cue words in two consecutive experimental runs. The representational similarity was then calculated between trials from different runs, both within the individual task stages (i.e., encoding–encoding similarity [EES], maintenance–maintenance similarity [MMS]) and across different task stages (i.e., encoding–maintenance similarity [EMS]). (B) Normalized electrode localization map. Each colored sphere indicates one channel, with different colors representing different participants. (C) Global RSA was performed by correlating iEEG power across 43 frequencies (between 2 and 100 Hz) and across all clean channels in consecutive time windows of 200 ms, sliding in steps of 10 ms. (Left) We correlated activity patterns during one trial from the first run (left, between dashed lines) with activity patterns during another trial in a second run (right, between dashed lines). (Right) Resulting similarity map across various time windows. The different sections (separated by dashed lines) show within- and across-stage similarity, corresponding to EES, EMS, and MMS in A.