Meeting Banner
Abstract #2331

Decoding Subjectively Correct "Yes/No" Thoughts in the Human Brain

Zhi Yang1, 2, Javier Gonzalez-Castillo2, Zirui Huang1, Rui Dai1, Georg Northoff3, Peter A. Bandettini2

1Institute of Psychology, Chinese Academy of Sciences, Beijing, China; 2National Institute of Mental Health, Bethesda, MD, United States; 3University of Otawa, Otawa, ON, Canada

Multivariate pattern analyses were used to decode the subjectively correct "Yes/No" answers to binary questions. Using a spatiotemporal searchlight approach, a set of brain regions were identified in 10 subjects in a 3T scanner, showing group-level above-chance accuracy in decoding "Yes/No" answers regardless the subjects' intentions that were manipulated in the experimental paradigm. The results from 7T scans further verified that three of these regions can be used to robustly decode the Yes/No answers (regardless of intentions) with high accuracy, given sufficiently high TSNR, which can be achieved by means of ultra-high field scanners and trial-averaging. These findings suggest that subjectively correct answers can be accurately decoded with fMRI in the spatial-temporal patterns of prefrontal cortex, providing a basis for fMRI-based brain-computer interface.

Abstract PDF

Keywords

absence academy accompany accuracy accurately achieved acquisition activity advance anatomical answers applied averaging awareness bands basic basis binary bold brain chance china classifier classifiers colored common communication compatible computer conducted confidence confirm consisted controls conversion correct corrected correction cortex cross dataset decided decode decoded decoding delay detected detecting direct display distinct encoding evaluate examining experiment extended field final findings frontal functional health honestly human include increasing indeed individual institute instruction insular intentions interface investigate knowledge labels leave marginal medial memories mental middle motion motor navigation needed nine noise notably onset panel panels paradigm patterns period permutation peter predict providing question questions random randomly reaching regardless regression respond response responses scanned scanner scheme science sciences searchlight semi separately signatures significantly simple spatial split subgroup subject subjectively subjects subsequently sufficient sufficiently suggest summarized superior table task tasks temporal thoughts trained training transparent trend trial trials ultrahigh validation vegetative verified visual vote whether yang