CUDA extension for structured kernels (Cauchy and Vandermonde multiplication) not found. Install by going to extensions/kernels/ and running `python setup.py install`, for improved speed and memory efficiency. Note that the kernel changed for state-spaces 4.0 and must be recompiled. Falling back on slow Cauchy and Vandermonde kernel. Install at least one of pykeops or the CUDA extension for better speed and memory efficiency. D. decision rule alone on pure-noise evidence (1500 trials of 60 s): antisymmetry and share of P > 0.5 by position rule antisym_err 0-5 s 5-30 s 30-60 s HMM 0.998 8.88e-16 0.488 0.509 0.491 B1 m20 k0.6 1.81e-14 0.485 0.493 0.499 B1 m10 k0.6 1.18e-14 0.498 0.498 0.501 B1 m5 k1.5 4.22e-15 0.508 0.503 0.5 D. OK B. column swap (streams exchanged): max |d' + d| / sd(d) and max |P' + P - 1|, sign agreement of d' = -d; 3 trials per dataset rel_evidence post_err sign_agree pipeline decoder (lookahead_s = 0.5) 0.00e+00 2.22e-16 1.0000 direct classifier 10 s, 3 blocks 0.00e+00 8.88e-16 1.0000 fusion a=0.5, DEC_OFFSET=3 (default) | HMM 0.998 0.00e+00 2.22e-16 1.0000 fusion a=0.5, DEC_OFFSET=3 (default) | B1 m20 k0.6 0.00e+00 1.18e-14 1.0000 fusion a=0.5, DEC_OFFSET=3 (default) | B1 m10 k0.6 0.00e+00 9.55e-15 1.0000 fusion a=0.5, DEC_OFFSET=3 (default) | B1 m5 k1.5 0.00e+00 2.89e-15 1.0000 B. column swap: EXACT FLIP (< 1e-3) C1. EEG with the stimuli of another trial: share of steps with d > 0 (column 0 favoured), by position in the trial; mean +- SE over pairs position 0-5 s 5-15 s 15-30 s 30-60 s pipeline pairs decoder (lookahead_s = 0.5) mismatched 0.488 0.501 0.499 0.507 matched 0.555 0.568 0.502 0.504 direct classifier 10 s, 3 blocks mismatched 0.455 0.474 0.533 0.505 matched 0.484 0.607 0.756 0.607 fusion a=0.5, DEC_OFFSET=3 (default) | HMM 0.998 mismatched 0.468 0.494 0.505 0.498 matched 0.532 0.616 0.622 0.555 C1. mismatched stimuli: largest |share - 0.5| = 0.045 -> NO POSITION / COLUMN PRIOR C2. trial without its first 20 s vs the full trial (evidence 30 s after the cut); correlation and sign agreement, 3 trials per dataset corr agree pipeline decoder (lookahead_s = 0.5) 0.996 0.970 direct classifier 10 s, 3 blocks 0.997 0.983 fusion a=0.5, DEC_OFFSET=3 (default) | HMM 0.998 0.996 0.973 A AVGC done A DTU done A KUL done A UHD done A USTC done A. lookahead by perturbation (5 datasets x 10 positions x 3 modes) pipeline perturbed tests changed min_s max_s verdict decoder (lookahead_s = 0.5) EEG 50 50 0.7500 0.9688 ok direct classifier 10 s, 3 blocks EEG 50 50 0.7500 0.9688 ok NEGATIVE CONTROL: decoder, offline read-out (3.75 s future) EEG 50 50 1.7500 3.7188 VIOLATION (> 1 s) fusion a=0.5, DEC_OFFSET=3 (default) | HMM 0.998 EEG 50 50 0.7500 0.9688 ok fusion a=0.5, DEC_OFFSET=3 (default) | B1 m20 k0.6 EEG 50 50 0.7500 0.9688 ok fusion a=0.5, DEC_OFFSET=3 (default) | B1 m10 k0.6 EEG 50 50 0.7500 0.9688 ok fusion a=0.5, DEC_OFFSET=3 (default) | B1 m5 k1.5 EEG 50 50 0.7500 0.9688 ok NEGATIVE CONTROL: fusion with DEC_OFFSET=2 (0.25 s too early) | HMM 0.998 EEG 50 50 1.0000 1.2188 VIOLATION (> 1 s) NEGATIVE CONTROL: fusion with DEC_OFFSET=2 (0.25 s too early) | B1 m20 k0.6 EEG 50 50 1.0000 1.2188 VIOLATION (> 1 s) NEGATIVE CONTROL: fusion with DEC_OFFSET=2 (0.25 s too early) | B1 m10 k0.6 EEG 50 50 1.0000 1.2188 VIOLATION (> 1 s) NEGATIVE CONTROL: fusion with DEC_OFFSET=2 (0.25 s too early) | B1 m5 k1.5 EEG 50 50 1.0000 1.2188 VIOLATION (> 1 s) NEGATIVE CONTROL: B1 filter run backwards in time EEG 50 50 13.9688 47.7188 VIOLATION (> 1 s) decoder (lookahead_s = 0.5) stimulus 50 50 0.2500 0.4688 ok direct classifier 10 s, 3 blocks stimulus 50 50 0.7500 0.9688 ok NEGATIVE CONTROL: decoder, offline read-out (3.75 s future) stimulus 50 50 0.2500 0.4688 ok fusion a=0.5, DEC_OFFSET=3 (default) | HMM 0.998 stimulus 50 50 0.7500 0.9688 ok fusion a=0.5, DEC_OFFSET=3 (default) | B1 m20 k0.6 stimulus 50 50 0.7500 0.9688 ok fusion a=0.5, DEC_OFFSET=3 (default) | B1 m10 k0.6 stimulus 50 50 0.7500 0.9688 ok fusion a=0.5, DEC_OFFSET=3 (default) | B1 m5 k1.5 stimulus 50 50 0.7500 0.9688 ok NEGATIVE CONTROL: fusion with DEC_OFFSET=2 (0.25 s too early) | HMM 0.998 stimulus 50 50 0.7500 0.9688 ok NEGATIVE CONTROL: fusion with DEC_OFFSET=2 (0.25 s too early) | B1 m20 k0.6 stimulus 50 50 0.7500 0.9688 ok NEGATIVE CONTROL: fusion with DEC_OFFSET=2 (0.25 s too early) | B1 m10 k0.6 stimulus 50 50 0.7500 0.9688 ok NEGATIVE CONTROL: fusion with DEC_OFFSET=2 (0.25 s too early) | B1 m5 k1.5 stimulus 50 50 0.7500 0.9688 ok NEGATIVE CONTROL: B1 filter run backwards in time stimulus 50 50 13.9688 47.7188 VIOLATION (> 1 s) decoder (lookahead_s = 0.5) EEG+stimulus 50 50 0.7500 0.9688 ok direct classifier 10 s, 3 blocks EEG+stimulus 50 50 0.7500 0.9688 ok NEGATIVE CONTROL: decoder, offline read-out (3.75 s future) EEG+stimulus 50 50 1.7500 3.7188 VIOLATION (> 1 s) fusion a=0.5, DEC_OFFSET=3 (default) | HMM 0.998 EEG+stimulus 50 50 0.7500 0.9688 ok fusion a=0.5, DEC_OFFSET=3 (default) | B1 m20 k0.6 EEG+stimulus 50 50 0.7500 0.9688 ok fusion a=0.5, DEC_OFFSET=3 (default) | B1 m10 k0.6 EEG+stimulus 50 50 0.7500 0.9688 ok fusion a=0.5, DEC_OFFSET=3 (default) | B1 m5 k1.5 EEG+stimulus 50 50 0.7500 0.9688 ok NEGATIVE CONTROL: fusion with DEC_OFFSET=2 (0.25 s too early) | HMM 0.998 EEG+stimulus 50 50 1.0000 1.2188 VIOLATION (> 1 s) NEGATIVE CONTROL: fusion with DEC_OFFSET=2 (0.25 s too early) | B1 m20 k0.6 EEG+stimulus 50 50 1.0000 1.2188 VIOLATION (> 1 s) NEGATIVE CONTROL: fusion with DEC_OFFSET=2 (0.25 s too early) | B1 m10 k0.6 EEG+stimulus 50 50 1.0000 1.2188 VIOLATION (> 1 s) NEGATIVE CONTROL: fusion with DEC_OFFSET=2 (0.25 s too early) | B1 m5 k1.5 EEG+stimulus 50 50 1.0000 1.2188 VIOLATION (> 1 s) NEGATIVE CONTROL: B1 filter run backwards in time EEG+stimulus 50 50 13.9688 47.7188 VIOLATION (> 1 s) A. submission pipelines: ALL WITHIN 1 s (largest lookahead used 0.9688 s); negative controls (EEG perturbed): all flagged SUMMARY: D ok, B ok, C ok, A ok